Modèles d’IA et modèles de langage

Comparez l’identité exacte des modèles, les entrées documentées, les limites de contexte et les tarifs par token. La documentation du fournisseur n’est pas un benchmark de qualité indépendant.

  • 20 outils comparés
  • Sources vérifiées jusqu’au 11 nov. 2026

Explorer les outils

20 fiches

Comparer deux ou trois outils côte à côte

Choisissez deux ou trois produits.

Comparer les preuves des sources

  • Documenté comme pris en charge
  • Documenté comme non pris en charge
  • Non vérifié Les sources ne l’établissent pas
  • Sans objet
Périmètre et limites documentés ; les critères non vérifiés restent inconnus.
ExigenceGPT-6.1 SolClaude Sonnet 5.5Gemini 3.8 FlashQwen3-8BOLMo-2-1124-7B-InstructDeepSeek V3Kimi K2GLM-4.5InternLM3IBM Granite 4.1 Language ModelsMicrosoft Phi-4gpt-ossQwen3-VLMiMo-V2-FlashStep 3.5 FlashLongCat-Flash-ChatMiniCPMSmolLMGLM-5 seriesIntern-S series
Tarifs
Tarification API à l’usage

Prix à revérifier

Tarification API à l’usage

Prix à revérifier

Tarification API à l’usage

Prix à revérifier

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Offre gratuite · prix payants non vérifiés

Pour qui
An OpenAI model accepting text and images and producing text.

Vérifié le 18 oct. 2026 · GPT-6.1 Sol: official API documentation

An Anthropic Claude model with text and image inputs and text output.

Vérifié le 18 oct. 2026 · Claude Sonnet 5.5: official API documentation

A Google Gemini model accepting text, images, video, audio and PDFs, with text output.

Vérifié le 18 oct. 2026 · Gemini 3.8 Flash: official API documentation

Qwen3-8B is documented as a causal language model with pretraining and post-training.

Vérifié le 3 nov. 2026 · Qwen3-8B: qwen3-8b-model-card

OLMo 2 7B Instruct is a post-trained variant of the OLMo 2 7B model.

Vérifié le 3 nov. 2026 · OLMo-2-1124-7B-Instruct: olmo2-7b-instruct-model-card

DeepSeek describes V3 as a mixture-of-experts language model with 671 billion total and 37 billion activated parameters.

Vérifié le 4 nov. 2026 · DeepSeek V3: deepseek-source

Moonshot AI describes Kimi K2 as a mixture-of-experts language model with 1 trillion total and 32 billion activated parameters.

Vérifié le 4 nov. 2026 · Kimi K2: kimi-source

The publisher describes GLM-4.5 as a foundation model with 355 billion total and 32 billion active parameters.

Vérifié le 4 nov. 2026 · GLM-4.5: glm-4-5-model-source

The publisher describes InternLM3-8B-Instruct as an 8-billion-parameter instruction model.

Vérifié le 4 nov. 2026 · InternLM3: internlm3-model-source

IBM describes Granite 4.1 language models as a dense decoder-only family with 3B, 8B and 30B sizes.

Publisher project README at a pinned commit; the sizes and base/instruct variants are not separately counted identities.

Vérifié le 4 nov. 2026 · IBM Granite 4.1 Language Models: granite-source

Microsoft describes Phi-4 as a 14B-parameter dense decoder-only model for text input.

Microsoft model card at a pinned revision; this is the named Phi-4 model, not an automatic claim about mini or multimodal variants.

Vérifié le 4 nov. 2026 · Microsoft Phi-4: phi-model-source

gpt-oss is a series of open-weight language models from OpenAI that the README describes as designed for reasoning, agentic tasks and developer use cases, released in two variants.

Vérifié le 8 nov. 2026 · gpt-oss: gpt-oss-project

Qwen3-VL is introduced in the README as a vision-language model in the Qwen series.

Vérifié le 8 nov. 2026 · Qwen3-VL: qwen3-vl-project

MiMo-V2-Flash is a Mixture-of-Experts language model from the XiaomiMiMo project with 309B total and 15B active parameters, designed for high-speed reasoning and agentic workflows according to its README.

Vérifié le 8 nov. 2026 · MiMo-V2-Flash: mimo-v2-flash-project

Step 3.5 Flash is an open-source foundation model aimed at reasoning and agentic tasks, built as a sparse Mixture of Experts that activates 11B of its 196B parameters per token.

Vérifié le 8 nov. 2026 · Step 3.5 Flash: step-3-5-flash-project

LongCat-Flash-Chat is a non-thinking foundation model from the Meituan LongCat team that its README says has particular strengths in agentic tasks.

Vérifié le 8 nov. 2026 · LongCat-Flash-Chat: longcat-flash-chat-project

MiniCPM5-1B, the first checkpoint of the MiniCPM5 series, is described in the README as a compact model for local assistants, coding agents, tool-use workflows and reasoning.

Vérifié le 8 nov. 2026 · MiniCPM: minicpm-project

The README describes SmolLM, presented as Smol Models, as a Hugging Face family of compact open models for text and vision that are meant to run on-device.

Vérifié le 8 nov. 2026 · SmolLM: smollm-project

The README presents GLM-5 as a model aimed at complex systems engineering and long-horizon agentic tasks.

Vérifié le 8 nov. 2026 · GLM-5 series: glm-5-project

The README introduces Intern-S2-397B as a multimodal foundation model for scientific intelligence and long-horizon agents.

Vérifié le 8 nov. 2026 · Intern-S series: intern-s1-project

Limites de couverture
The 1,050,000-token context window and 128,000-token output ceiling are different limits. Long prompts and premium service tiers change token rates.

Vérifié le 18 oct. 2026 · GPT-6.1 Sol: official API documentation

The standard output ceiling is 128K tokens; the documented 300K batch option requires a separate beta header. Thinking and tool behavior are model-specific.

Vérifié le 18 oct. 2026 · Claude Sonnet 5.5: official API documentation

The documented input limit is 1,048,576 tokens and output limit is 65,536. The standard paid rates shown here expire after 31 December 2026.

Vérifié le 18 oct. 2026 · Gemini 3.8 Flash: official API documentation

The model card lists a native 32,768-token context and a longer context with YaRN.

Vérifié le 3 nov. 2026 · Qwen3-8B: qwen3-8b-model-card

The model card describes the model language as primarily English.

Vérifié le 3 nov. 2026 · OLMo-2-1124-7B-Instruct: olmo2-7b-instruct-model-card

The project states that its code and model weights are governed by separate licences.

Vérifié le 4 nov. 2026 · DeepSeek V3: deepseek-source

The project identifies a Modified MIT License for code and model weights.

Vérifié le 4 nov. 2026 · Kimi K2: kimi-source

The captured GLM-4.5 project describes thinking and non-thinking modes.

Vérifié le 4 nov. 2026 · GLM-4.5: glm-4-5-model-source

The captured project gives an approximately 8 GB GPU-memory estimate for one 4-bit InternLM3 8B example.

Vérifié le 4 nov. 2026 · InternLM3: internlm3-model-source

IBM describes both base and instruction-tuned models in the Granite 4.1 family.

Publisher project README at a pinned commit; select the exact checkpoint, context configuration and runtime before deployment.

Vérifié le 4 nov. 2026 · IBM Granite 4.1 Language Models: granite-source

The Phi-4 model card specifies a 16K-token context length.

Microsoft model card at a pinned revision; verify the selected checkpoint and runtime configuration.

Vérifié le 4 nov. 2026 · Microsoft Phi-4: phi-model-source

Per the README, both gpt-oss models were trained with the harmony response format and should only be used with that format, otherwise they will not work correctly.

Vérifié le 8 nov. 2026 · gpt-oss: gpt-oss-project

The README states that Qwen3-VL has a native 256K context that is expandable to 1M.

Vérifié le 8 nov. 2026 · Qwen3-VL: qwen3-vl-project

The README states that the MiMo-V2-Flash context window supports lengths of up to 256k.

Vérifié le 8 nov. 2026 · MiMo-V2-Flash: mimo-v2-flash-project

The README says Step 3.5 Flash is tailored for coding and work-centric tasks and may be less stable in specialized domains or long multi-turn dialogues, with repetitive reasoning or mixed-language outputs.

Vérifié le 8 nov. 2026 · Step 3.5 Flash: step-3-5-flash-project

The README says the project has implemented basic adaptations in SGLang and vLLM to support deploying LongCat-Flash.

Vérifié le 8 nov. 2026 · LongCat-Flash-Chat: longcat-flash-chat-project

The README describes MiniCPM5-1B as a dense 1B-parameter Transformer built for on-device, local and resource-constrained deployment.

Vérifié le 8 nov. 2026 · MiniCPM: minicpm-project

The README states that SmolLM3 supports six languages: English, French, Spanish, German, Italian and Portuguese.

Vérifié le 8 nov. 2026 · SmolLM: smollm-project

Within the series, the README gives a 1M-token context for GLM-5.2.

Vérifié le 8 nov. 2026 · GLM-5 series: glm-5-project

The README lists LMDeploy, vLLM and SGLang as the inference frameworks for deploying the Intern-S2 and Intern-S1 series models.

Vérifié le 8 nov. 2026 · Intern-S series: intern-s1-project

Fournisseur
OpenAI

Provider identity is linked to the separate OpenAI API catalogue entry.

Vérifié le 18 oct. 2026 · GPT-6.1 Sol: official API documentation

Anthropic

Provider identity is linked to the separate Anthropic Claude API entry.

Vérifié le 18 oct. 2026 · Claude Sonnet 5.5: official API documentation

Google

Provider identity is linked to the separate Google Gemini API entry.

Vérifié le 18 oct. 2026 · Gemini 3.8 Flash: official API documentation

Qwen Team

Publisher model card (Hugging Face README at pinned revision b968826d) captured 9 November 2026; supplier description, not a deployment or independent test. The card credits the Qwen Team as author of the cited Qwen3 technical report; it does not describe a hosted service or commercial provider.

Vérifié le 10 nov. 2026 · Qwen3-8B: qwen3-8b-modelcard-20261109b

Allen Institute for AI

Publisher model card (Hugging Face README at pinned revision 470b1fba) captured 9 November 2026; supplier description, not a deployment or independent test. The card’s example system prompt says the assistant is built by the Allen Institute for AI; the card does not describe a hosted API.

Vérifié le 10 nov. 2026 · OLMo-2-1124-7B-Instruct: olmo2-7b-instruct-modelcard-20261109b

DeepSeek

Provider named by the DeepSeek License Agreement 1.0 definition of "DeepSeek" (LICENSE-MODEL at commit 9b4e978, captured 9 November 2026); the legal entities are Beijing and Hangzhou DeepSeek Artificial Intelligence Fundamental Technology Research Co., Ltd. and affiliates.

Vérifié le 9 nov. 2026 · DeepSeek V3: deepseek-v3-modellicense-project

Moonshot AI

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. The README links the publisher homepage under the name Moonshot AI; no hosted service or account was used.

Vérifié le 10 nov. 2026 · Kimi K2: kimi-k2-project

Non vérifié
Shanghai AI Laboratory

Publisher project README at the pinned commit, captured on 9 November 2026. The README names the publisher only through the ModelScope namespace Shanghai_AI_Laboratory in the Model Zoo link; no hosted service or account was used.

Vérifié le 10 nov. 2026 · InternLM3: internlm3-model-project

IBM Research

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. Publisher named in the README citation block (author IBM Research); no hosted service or account was used.

Vérifié le 10 nov. 2026 · IBM Granite 4.1 Language Models: ibm-granite-4-1-project

Microsoft Research

Publisher model card (Hugging Face README at pinned revision 2db69c1c) captured 9 November 2026; supplier description, not a deployment or independent test. The card names Microsoft Research as developer of the open model; it does not describe a hosted service.

Vérifié le 10 nov. 2026 · Microsoft Phi-4: microsoft-phi-4-modelcard-20261109b

OpenAI

Publisher project README at tag v0.0.9 (commit 5994767) captured 9 November 2026; supplier description, not a deployment, inference or independent test. The README calls gpt-oss OpenAI's open-weight models.

Vérifié le 10 nov. 2026 · gpt-oss: gpt-oss-project

Non vérifié
Xiaomi

Dated publisher project README (GitHub, commit 941e5547) captured 9 November 2026; supplier description, not a deployment or independent test. The README’s recommended system prompt says the assistant is developed by Xiaomi; hosted-service terms were not captured.

Vérifié le 10 nov. 2026 · MiMo-V2-Flash: mimo-v2-flash-project

Non vérifié
Meituan LongCat Team

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. Author named in the README citation block; no hosted service or account was used.

Vérifié le 10 nov. 2026 · LongCat-Flash-Chat: longcat-flash-chat-project

Modelbest Inc., THUNLP and the Gaoling School of Artificial Intelligence of RUC

Dated publisher project README (GitHub release 5.0, commit b4a5d6cd) captured 9 November 2026; supplier description, not a deployment or independent test. The README lists these institutions as developers of the project; it does not describe a hosted service.

Vérifié le 10 nov. 2026 · MiniCPM: minicpm-project

Hugging Face

Dated publisher project README captured on 9 November 2026 (commit 438be18d); names the publisher of the open models, not a hosted service.

Vérifié le 11 nov. 2026 · SmolLM: smollm-project

Z.ai

Publisher project README at commit 83c03d3 captured 9 November 2026; supplier description, not a deployment, inference or independent test. The README points to the Z.ai API Platform for GLM-5.3 and GLM-5.3-Flash API services; the licence copyright line reads Z.AI.

Vérifié le 10 nov. 2026 · GLM-5 series: glm-5-series-project

Non vérifié
Identifiant du modèle dans l’API
gpt-6.1-sol

Vérifié le 18 oct. 2026 · GPT-6.1 Sol: official API documentation

claude-sonnet-5-5

Vérifié le 18 oct. 2026 · Claude Sonnet 5.5: official API documentation

gemini-3.8-flash

Vérifié le 18 oct. 2026 · Gemini 3.8 Flash: official API documentation

Qwen/Qwen3-8B

Publisher model card (Hugging Face README at pinned revision b968826d) captured 9 November 2026; supplier description, not a deployment or independent test. Hugging Face repository identifier used in the card’s code sample, not the name of a hosted API model; no code was run.

Vérifié le 10 nov. 2026 · Qwen3-8B: qwen3-8b-modelcard-20261109b

allenai/OLMo-2-1124-7B-Instruct

Publisher model card (Hugging Face README at pinned revision 470b1fba) captured 9 November 2026; supplier description, not a deployment or independent test. Hugging Face repository identifier used in the card’s code sample, not the name of a hosted API model; no code was run.

Vérifié le 10 nov. 2026 · OLMo-2-1124-7B-Instruct: olmo2-7b-instruct-modelcard-20261109b

DeepSeek-V3

Vérifié le 4 nov. 2026 · DeepSeek V3: deepseek-source

Kimi-K2-Instruct

Vérifié le 4 nov. 2026 · Kimi K2: kimi-source

zai-org/GLM-4.5

Publisher project README at commit 170f20b captured 9 November 2026; supplier description, not a deployment, inference or independent test. Hugging Face repository name of the GLM-4.5 weights in the README's download table; FP8, Air and Base variants have their own repository names.

Vérifié le 10 nov. 2026 · GLM-4.5: glm-4-5-model-project

InternLM3-8B-Instruct, internlm/internlm3-8b-instruct

Publisher project README at the pinned commit, captured on 9 November 2026. Model Zoo entry; the same README also lists older InternLM2 and InternLM2.5 models, which are not recorded here. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · InternLM3: internlm3-model-project

ibm-granite/granite-4.1-3b, ibm-granite/granite-4.1-3b-base, ibm-granite/granite-4.1-8b, ibm-granite/granite-4.1-8b-base, ibm-granite/granite-4.1-30b, ibm-granite/granite-4.1-30b-base

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. The sizes and base/instruct variants are checkpoints of one family, not separately counted products. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · IBM Granite 4.1 Language Models: ibm-granite-4-1-project

microsoft/phi-4

Publisher model card (Hugging Face README at pinned revision 2db69c1c) captured 9 November 2026; supplier description, not a deployment or independent test. Hugging Face repository identifier used in the card’s code sample, not the name of a hosted API model; no code was run.

Vérifié le 10 nov. 2026 · Microsoft Phi-4: microsoft-phi-4-modelcard-20261109b

gpt-oss-120b and gpt-oss-20b

Publisher project README at tag v0.0.9 (commit 5994767) captured 9 November 2026; supplier description, not a deployment, inference or independent test. The two released models; Hugging Face repositories are openai/gpt-oss-120b and openai/gpt-oss-20b.

Vérifié le 10 nov. 2026 · gpt-oss: gpt-oss-project

Qwen/Qwen3-VL-235B-A22B-Instruct

Dated publisher project README captured on 9 November 2026 (commit 96588727); supplier description, not a deployment or independent test. Hugging Face repository ID of the largest Instruct model the README lists. The README also lists 2B, 4B, 8B, 30B-A3B and 32B variants with Instruct and Thinking editions; a named variant is evidence within the model family, not a separate product.

Vérifié le 10 nov. 2026 · Qwen3-VL: qwen3-vl-project

XiaomiMiMo/MiMo-V2-Flash

Dated publisher project README (GitHub, commit 941e5547) captured 9 November 2026; supplier description, not a deployment or independent test. Hugging Face repository identifier used in the README’s SGLang launch command, not the name of a hosted API model; nothing was deployed.

Vérifié le 10 nov. 2026 · MiMo-V2-Flash: mimo-v2-flash-project

step-3.5-flash

Dated publisher project README captured on 9 November 2026 (commit 21d85a5f); supplier description, not a deployment or independent test. API model ID used in the README's agent-platform guides; OpenRouter uses stepfun/step-3.5-flash.

Vérifié le 11 nov. 2026 · Step 3.5 Flash: step-3-5-flash-project

LongCat-Flash-Chat

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. Model name as released in the README. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · LongCat-Flash-Chat: longcat-flash-chat-project

openbmb/MiniCPM5-1B

Dated publisher project README (GitHub release 5.0, commit b4a5d6cd) captured 9 November 2026; supplier description, not a deployment or independent test. Hugging Face repository identifier of the current release used in the README’s code sample, not the name of a hosted API model; no code was run.

Vérifié le 10 nov. 2026 · MiniCPM: minicpm-project

HuggingFaceTB/SmolLM3-3B

Model card of HuggingFaceTB/SmolLM3-3B on Hugging Face, captured 9 November 2026; supplier description, not a deployment or test. Hugging Face repository ID of the SmolLM3 3B instruct model; a named variant is evidence within the model family, not a separate product.

Vérifié le 11 nov. 2026 · SmolLM: smollm-hf-smollm3-3b-20261109b

GLM-5.3, GLM-5.2, GLM-5.1, GLM-5

Publisher project README at commit 83c03d3 captured 9 November 2026; supplier description, not a deployment, inference or independent test. Names in the README title; GLM-5.3-Flash and FP8/BF16 builds have their own repository names in the download table.

Vérifié le 10 nov. 2026 · GLM-5 series: glm-5-series-project

internlm/Intern-S2-397B, internlm/Intern-S2-397B-FP8

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. The README Model Zoo lists six model groups (Intern-S2-397B, Intern-S2-Preview-397B, Intern-S2-Preview-35B, Intern-S1-Pro, Intern-S1, Intern-S1-mini); only the Intern-S2-397B repositories (BF16 and FP8) are recorded here. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · Intern-S series: intern-s1-project

Modalités
Text and image input; text output

Vérifié le 18 oct. 2026 · GPT-6.1 Sol: official API documentation

Text and image input; text output

Vérifié le 18 oct. 2026 · Claude Sonnet 5.5: official API documentation

Text, image, video, audio and PDF input; text output

Vérifié le 18 oct. 2026 · Gemini 3.8 Flash: official API documentation

Text generation

Vérifié le 3 nov. 2026 · Qwen3-8B: qwen3-8b-model-card

Text generation

Vérifié le 3 nov. 2026 · OLMo-2-1124-7B-Instruct: olmo2-7b-instruct-model-card

Non vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifié
Text input (best suited for chat-format prompts); generated text output

Publisher model card (Hugging Face README at pinned revision 2db69c1c) captured 9 November 2026; supplier description, not a deployment or independent test. Supplier description of inputs and outputs; no multimodal input is described.

Vérifié le 10 nov. 2026 · Microsoft Phi-4: microsoft-phi-4-modelcard-20261109b

Non vérifié
Image and text input; text output

Hugging Face pipeline tag on the Qwen3-VL-8B-Instruct model card, captured 9 November 2026; a task label, not a tested modality matrix. The project README also describes video understanding.

Vérifié le 10 nov. 2026 · Qwen3-VL: qwen3-vl-hf-qwen3-vl-8b-instruct-20261109b

Non vérifiéNon vérifiéNon vérifiéNon vérifié
Text language model

Model card of HuggingFaceTB/SmolLM3-3B on Hugging Face, captured 9 November 2026; supplier description, not a deployment or test. Scope of the SmolLM3 language model only; the family's SmolVLM vision model is documented separately in the project README.

Vérifié le 11 nov. 2026 · SmolLM: smollm-hf-smollm3-3b-20261109b

Non vérifié
Multimodal (vision-language)

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. The README describes Intern-S2-397B as multimodal with vision-language pre-training; it does not itemise the accepted input and output types in this section, and other series members differ. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · Intern-S series: intern-s1-project

Fenêtre de contexte
1 050 000 tokens in the context window

Vérifié le 18 oct. 2026 · GPT-6.1 Sol: official API documentation

1M tokens in the context window

Vérifié le 18 oct. 2026 · Claude Sonnet 5.5: official API documentation

1 048 576 input tokens (not a shared context-window measurement)

Vérifié le 18 oct. 2026 · Gemini 3.8 Flash: official API documentation

32,768 natively; 131,072 with YaRN tokens in the context window

Publisher model card (Hugging Face README at pinned revision b968826d) captured 9 November 2026; supplier description, not a deployment or independent test. Longer contexts need the YaRN rope-scaling setting that the card says can affect performance on shorter texts; no long-context test was run.

Vérifié le 10 nov. 2026 · Qwen3-8B: qwen3-8b-modelcard-20261109b

Non vérifié
128 000 tokens in the context window

Publisher repository README at commit 9b4e978 captured 9 November 2026; supplier description, not a deployment, inference or independent test. Context length listed for DeepSeek-V3 and DeepSeek-V3-Base in the Model Downloads table; the README's needle-in-a-haystack tests cover context lengths up to 128K. The hosted service may differ.

Vérifié le 9 nov. 2026 · DeepSeek V3: deepseek-v3-project

128K tokens in the context window

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. Context length from the README model summary table, which describes the model architecture; a hosted service may apply other limits. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · Kimi K2: kimi-k2-project

128 000 tokens in the context window

Publisher project README at commit 170f20b captured 9 November 2026; supplier description, not a deployment, inference or independent test. The README says the GLM-4.5 models can use their full 128K context length on the larger GPU configurations it lists; GLM-4.6 expands the context to 200K tokens and is a different model. Hosted API limits may differ.

Vérifié le 10 nov. 2026 · GLM-4.5: glm-4-5-model-project

Non vérifié
Up to 512K tokens in the context window

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. The README describes the final training phase as scaling the context window up to 512K tokens; the supported length of each 3B, 8B and 30B checkpoint and runtime configuration is not itemised, so check the model card before relying on it. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · IBM Granite 4.1 Language Models: ibm-granite-4-1-project

16K tokens in the context window

Publisher model card (Hugging Face README at pinned revision 2db69c1c) captured 9 November 2026; supplier description, not a deployment or independent test. Supplier specification; no long-context test was run.

Vérifié le 10 nov. 2026 · Microsoft Phi-4: microsoft-phi-4-modelcard-20261109b

Non vérifié
256K tokens (configured default; YaRN extension documented)

Dated publisher project README captured on 9 November 2026 (commit 96588727); supplier description, not a deployment or independent test. The default config.json is set for up to 256K tokens; going beyond that needs YaRN settings in config.json or in the vLLM serving arguments. Memory needs, speed and output quality at long lengths were not tested.

Vérifié le 10 nov. 2026 · Qwen3-VL: qwen3-vl-project

256k tokens in the context window

Dated publisher project README (GitHub, commit 941e5547) captured 9 November 2026; supplier description, not a deployment or independent test. The README states the length without naming the unit; its launch example sets a context length of 262144. No long-context test was run.

Vérifié le 10 nov. 2026 · MiMo-V2-Flash: mimo-v2-flash-project

256K tokens in the context window

Dated publisher project README captured on 9 November 2026 (commit 21d85a5f); supplier description, not a deployment or independent test. Supplier statement of the supported context window; no long-context test was run.

Vérifié le 11 nov. 2026 · Step 3.5 Flash: step-3-5-flash-project

128k tokens in the context window

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. The README says mid-training extended the context length to 128k for agentic post-training; it does not state a separate supported maximum for a deployment, so treat this as the documented training extension, not a guaranteed limit. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · LongCat-Flash-Chat: longcat-flash-chat-project

Non vérifié
128K tokens (up to, with YaRN extrapolation)

Model card of HuggingFaceTB/SmolLM3-3B on Hugging Face, captured 9 November 2026; supplier description, not a deployment or test. The card says the model was trained on 64k context and the shipped config.json is set for up to 65,536 tokens; reaching 128K needs the documented YaRN rope-scaling change. Quality at long lengths was not tested.

Vérifié le 11 nov. 2026 · SmolLM: smollm-hf-smollm3-3b-20261109b

Non vérifiéNon vérifié
Sortie maximale
128 000 output tokens

Vérifié le 18 oct. 2026 · GPT-6.1 Sol: official API documentation

128K output tokens (standard API)

Vérifié le 18 oct. 2026 · Claude Sonnet 5.5: official API documentation

65 536 output tokens

Vérifié le 18 oct. 2026 · Gemini 3.8 Flash: official API documentation

Non vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifié
Tarif par token d’entrée
2 USD per million input tokens

Standard uncached text input for prompts up to 272K input tokens. Longer prompts: 2x input/cache and 1.5x output for the full request. Fast 2x, Ultrafast 6x; regional processing adds 10% where available. Cache writes, tools and tax are additional scopes.

Vérifié le 18 oct. 2026 · GPT-6.1 Sol: official API documentation

2 USD per million base input tokens

Standard base input rate for Claude Sonnet 5.5. Prompt cache writes/reads and batch processing have separate rates; tools, service tier and taxes can change the total.

Vérifié le 18 oct. 2026 · Claude Sonnet 5.5: official API documentation

0,75 USD per million input tokens

Gemini 3.8 Flash Standard paid tier through 31 December 2026; USD 1.50 from 1 January 2027. Cache storage, grounding, other service tiers and taxes are separate.

Vérifié le 18 oct. 2026 · Gemini 3.8 Flash: official API documentation

Non vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifié
Tarif par token de sortie
10 USD per million output tokens

Standard text output for prompts up to 272K input tokens. Above 272K input tokens: 1.5x output rate. Fast, Ultrafast, regional processing, tools and tax can change the total.

Vérifié le 18 oct. 2026 · GPT-6.1 Sol: official API documentation

10 USD per million output tokens

Standard output rate for Claude Sonnet 5.5. Compare the exact consumption option; the batch discount is separate.

Vérifié le 18 oct. 2026 · Claude Sonnet 5.5: official API documentation

3,75 USD per million output tokens, including thinking

Gemini 3.8 Flash Standard paid tier through 31 December 2026; USD 7.50 from 1 January 2027. Other service tiers, cache storage, grounding and taxes are separate.

Vérifié le 18 oct. 2026 · Gemini 3.8 Flash: official API documentation

Non vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifiéNon vérifié
Disponibilité
Listed for paid API usage tiers; Free tier is not supported.

Published rate-limit table, not a verification of regional eligibility or access for a particular account.

Vérifié le 18 oct. 2026 · GPT-6.1 Sol: official API documentation

Active model; released 28 September 2026.

Vendor lifecycle status checked on 18 October 2026; account and platform eligibility require confirmation.

Vérifié le 18 oct. 2026 · Claude Sonnet 5.5: official API documentation

Stable model alias; documentation lists a September 2026 update.

Documentation status, not an account-specific access guarantee.

Vérifié le 18 oct. 2026 · Gemini 3.8 Flash: official API documentation

Non vérifié
The model card notes a 1/3/2025 update: because the first post-trained models did not share the base models’ pre-tokenization logic, new post-trained models were trained and published under the same names, and the old ones carry a "-preview" postfix.

Publisher model card (Hugging Face README at pinned revision 470b1fba) captured 9 November 2026; supplier description, not a deployment or independent test. The card writes the update date as 1/3/2025 without stating the date order. The note concerns which checkpoint the name points to; confirm the revision before relying on a result.

Vérifié le 10 nov. 2026 · OLMo-2-1124-7B-Instruct: olmo2-7b-instruct-modelcard-20261109b

Available on DeepSeek's official chat website and through an OpenAI-compatible API on the DeepSeek Platform.

Publisher repository README at commit 9b4e978 captured 9 November 2026; supplier description, not a deployment, inference or independent test. The README does not say which model version those hosted services currently run, and their prices and limits are not captured here; open weights are covered by the separate Self-hosted plan facts. Account, region and platform eligibility require confirmation.

Vérifié le 9 nov. 2026 · DeepSeek V3: deepseek-v3-project

Block-fp8 checkpoints of Kimi-K2-Instruct are published on Hugging Face.

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. Checkpoint availability only. The README also points to a hosted API at platform.moonshot.ai; that service, its current model lineup, prices and terms were not verified here. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · Kimi K2: kimi-k2-project

GLM-4.5 (355B-A32B, BF16) open weights are listed on Hugging Face and ModelScope.

Publisher project README at commit 170f20b captured 9 November 2026; supplier description, not a deployment, inference or independent test. The README's download table also has separate entries for the FP8, Air and Base variants. Download listing, not a statement of current hosted availability; the README also points to the Z.ai API Platform for GLM-4.7 API services. Licence is covered by the separate pricing fact.

Vérifié le 10 nov. 2026 · GLM-4.5: glm-4-5-model-project

The InternLM3-8B-Instruct weights are listed on Hugging Face, ModelScope and Modelers, with a release date of 2025-01-15.

Publisher project README at the pinned commit, captured on 9 November 2026. Model Zoo table only; hosted chat or API access, its terms and prices were not verified here. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · InternLM3: internlm3-model-project

Open checkpoints are published on Hugging Face (ibm-granite/granite-4.1-* repositories).

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. Checkpoint availability only; the README example clones the 8B model from Hugging Face. Hosted offerings, their regions and prices were not verified here. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · IBM Granite 4.1 Language Models: ibm-granite-4-1-project

Static model released on 12 December 2024, trained on an offline dataset with data cutoffs of June 2024 and earlier for publicly available data.

Publisher model card (Hugging Face README at pinned revision 2db69c1c) captured 9 November 2026; supplier description, not a deployment or independent test. Release and cutoff dates are supplier statements; later revisions of the repository or hosted offerings were not checked.

Vérifié le 10 nov. 2026 · Microsoft Phi-4: microsoft-phi-4-modelcard-20261109b

Model weights can be downloaded from the Hugging Face Hub (openai/gpt-oss-120b and openai/gpt-oss-20b).

Publisher project README at tag v0.0.9 (commit 5994767) captured 9 November 2026; supplier description, not a deployment, inference or independent test. Download instruction, not a statement of current hosted availability or of any OpenAI-hosted service. Licence is covered by the separate pricing fact.

Vérifié le 10 nov. 2026 · gpt-oss: gpt-oss-project

Self-hosted deployment is recommended with vLLM 0.11.0 or newer; an official Docker image is provided.

Dated publisher project README captured on 9 November 2026 (commit 96588727); supplier description, not a deployment or independent test. Deployment guidance for the weights, not a hosted-service entitlement; GPU requirements and costs depend on the chosen model size.

Vérifié le 10 nov. 2026 · Qwen3-VL: qwen3-vl-project

Non vérifié
Local inference is documented with vLLM, SGLang, Hugging Face Transformers and llama.cpp.

Dated publisher project README captured on 9 November 2026 (commit 21d85a5f); supplier description, not a deployment or independent test. Documentation of local backends, not a performance claim or a hosted-service entitlement; cloud API access through OpenRouter and the StepFun platform is described separately.

Vérifié le 11 nov. 2026 · Step 3.5 Flash: step-3-5-flash-project

A chat website is offered at https://longcat.ai.

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. The README invites readers to chat on the official website. Access terms, limits, cost and API availability of that site were not verified here. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · LongCat-Flash-Chat: longcat-flash-chat-project

The README changelog dates the release of MiniCPM5-1B to 19 May 2026.

Dated publisher project README (GitHub release 5.0, commit b4a5d6cd) captured 9 November 2026; supplier description, not a deployment or independent test. The changelog writes the date as 2026.05.19 and the README's download section labels MiniCPM5-1B the current release; later repository changes and model-hub revisions were not checked.

Vérifié le 10 nov. 2026 · MiniCPM: minicpm-project

Local inference is documented with llama.cpp, ONNX, MLX, MLC and ExecuTorch, with quantized checkpoints in the publisher's collection.

Model card of HuggingFaceTB/SmolLM3-3B on Hugging Face, captured 9 November 2026; supplier description, not a deployment or test. Documentation of local runtimes, not a performance claim or a hosted-service entitlement; the card also documents OpenAI-compatible serving with vLLM and SGLang.

Vérifié le 11 nov. 2026 · SmolLM: smollm-hf-smollm3-3b-20261109b

Open weights for GLM-5.3 (744B-A40B, FP8 and BF16) and GLM-5.3-Flash (320B-A18B, FP8) are listed for download on Hugging Face and ModelScope.

Publisher project README at commit 83c03d3 captured 9 November 2026; supplier description, not a deployment, inference or independent test. The README's table also lists GLM-5.3-Flash-BF16 and the GLM-5.2, GLM-5.1 and GLM-5 weights in FP8 and BF16. Download listing, not a statement of current hosted availability; the README also points to the Z.ai API Platform. Licences are covered by the separate pricing fact.

Vérifié le 10 nov. 2026 · GLM-5 series: glm-5-series-project

Model collections are published on Hugging Face and ModelScope, and an online chat is linked at chat.intern-ai.org.cn.

Publisher repository README at the pinned commit, captured on 9 November 2026; supplier description, not a deployment or independent test. Links at the top of the README. Online chat access terms, limits and cost were not verified here. No model inference, API call, account, benchmark or performance test was performed.

Vérifié le 10 nov. 2026 · Intern-S series: intern-s1-project

Note de couverture

Official documentation checked on 18 October 2026; no hands-on test.

Official documentation checked on 18 October 2026; no hands-on test.

Official documentation checked on 18 October 2026; no hands-on test.

Publisher-maintained model card captured 3 November 2026. Exact artifact scope; model performance, hosting availability, prices, privacy and safety not independently tested. No model weights, paid AI or inference request were used.

Publisher-maintained model card captured 3 November 2026. Exact artifact scope; model performance, hosting availability, prices, privacy and safety not independently tested. No model weights, paid AI or inference request were used.

Publisher repository README at the captured revision, 4 November 2026. A named variant is evidence within the model family, not a separate product. No model inference, API call, account, benchmark or performance test. Current availability and terms require separate review.

Publisher repository README at the captured revision, 4 November 2026. A named variant is evidence within the model family, not a separate product. No model inference, API call, account, benchmark or performance test. Current availability and terms require separate review.

Publisher project README at captured revision, 4 November 2026. Model and configuration specific; source estimates are not independent tests or universal deployment guarantees. Separate model licence page unavailable.

Publisher project README at captured revision, 4 November 2026. Model and configuration specific; source estimates are not independent tests or universal deployment guarantees. Separate model licence page unavailable.

Dated publisher project README captured on 7 November 2026; supplier description, not a deployment or independent test. Infrastructure, accounts, third-party services and managed offerings have separate costs and terms.

Dated publisher project README captured on 7 November 2026; supplier description, not a deployment or independent test. Infrastructure, accounts, third-party services and managed offerings have separate costs and terms.

Dated publisher project README captured on 7 November 2026; supplier description, not a deployment or independent test. Infrastructure, accounts, third-party services and managed offerings have separate costs and terms.

Dated publisher project README captured on 7 November 2026; supplier description, not a deployment or independent test. Infrastructure, accounts, third-party services and managed offerings have separate costs and terms.

Dated publisher project README captured on 7 November 2026; supplier description, not a deployment or independent test. Infrastructure, accounts, third-party services and managed offerings have separate costs and terms.

Dated publisher project README captured on 7 November 2026; supplier description, not a deployment or independent test. Infrastructure, accounts, third-party services and managed offerings have separate costs and terms.

Dated publisher project README captured on 7 November 2026; supplier description, not a deployment or independent test. Infrastructure, accounts, third-party services and managed offerings have separate costs and terms.

Dated publisher project README captured on 7 November 2026; supplier description, not a deployment or independent test. Infrastructure, accounts, third-party services and managed offerings have separate costs and terms.

Dated publisher project README captured on 7 November 2026; supplier description, not a deployment or independent test. Infrastructure, accounts, third-party services and managed offerings have separate costs and terms.

Face à face

Comparatifs mis en avant : modèles d’IA

Tous les comparatifs

Réponses rapides

Questions sur Modèles d’IA

Quels outils de modèles d’IA proposent une offre gratuite ?

Qwen3-8B, OLMo-2-1124-7B-Instruct, DeepSeek V3, Kimi K2, GLM-4.5, InternLM3, IBM Granite 4.1 Language Models, Microsoft Phi-4, gpt-oss, Qwen3-VL, MiMo-V2-Flash, Step 3.5 Flash, LongCat-Flash-Chat, MiniCPM, SmolLM, GLM-5 series et Intern-S series documentent une offre gratuite sur leurs pages de tarifs officielles. Les offres gratuites limitent généralement l’usage, le nombre d’utilisateurs ou les fonctionnalités ; vérifiez les limites dans chaque fiche.