Mistral-Medium-3.5-128B
Support summary
Mistral-Medium-3.5-128B is a checkpoint published by Mistral AI, built on the Mistral3ForConditionalGeneration architecture. Official recipes or cookbooks cover vLLM and SGLang, without pinning a stable release. TensorRT-LLM lists its architecture, but not this exact checkpoint.
- Documented hardware
- AMD / ROCm
- Documented quantization
- FP8
- Documented features
- Reasoning parsers, Speculative decoding, Tensor parallelism, and Tool calling
- Configured context
- 262,144 tokens
- License
- other
Hardware, quantization and features count only statements about this checkpoint; engine-wide capabilities are listed with each release.
Hugging Face model card
- Publisher
- Mistral AI
- License
- other
- Architecture
- Mistral3ForConditionalGeneration
- Weight tensor elements
- 127.7B
- Tensor types
- BF16, F8_E4M3
- Quantization
- fp8
- Configured context
- 262,144 tokens
- Languages
- en, fr, de, es, pt, it, ja, ko, ru, zh, ar, fa, id, ms, ne, pl, ro, sr, sv, tr, uk, vi, hi, bn
- Weight access
- Public
- Repository created
- 2026-03-31
Publisher deployment guidance
- vLLMnightly
The publisher uses vLLM nightly with mistral_common>=1.11.1 and transformers>=5.4.0.
Source: mistralai/Mistral-Medium-3.5-128B · publisher model cardReviewed - SGLangdevelopment containers
The publisher specifies dev-mistral-medium-3.5 for H100/H200 with CUDA 12.9, or dev-cu13-mistral-medium-3.5 for B200/B300 with CUDA 13.0. These are container builds, not host-driver minimums.
Source: mistralai/Mistral-Medium-3.5-128B · publisher model cardReviewed
Model notes and sources
- mistralai/Mistral-Medium-3.5-128B · publisher model card
- mistralai/Mistral-Medium-3.5-128B · Hugging Face metadata
- mistralai/Mistral-Medium-3.5-128B · publisher configuration
- vllm · models/mistralai/Mistral-Medium-3.5-128B.yaml
- sglang · docs/src/snippets/autoregressive/mistral-medium-3-5-deployment.jsx
- sglang · docs/cookbook/autoregressive/Mistral/Mistral-Medium-3.5.mdx
- Reviewed
Engine support
3 engines
| Engine | Release | Checkpoint support | Evidence scope | Last known supported |
|---|---|---|---|---|
| vLLM | Version not pinnedRecipe / cookbook | Documented recipeSource: vllm · models/mistralai/Mistral-Medium-3.5-128B.yamlReviewed | No supported release indexed | |
Features, dependencies and evidence for vLLM 0.28.0 | ||||
| SGLang | Version not pinnedRecipe / cookbook | Documented recipeSource 1: sglang · docs/cookbook/autoregressive/Mistral/Mistral-Medium-3.5.mdxSource 2: sglang · docs/src/snippets/autoregressive/mistral-medium-3-5-deployment.jsxReviewed | No supported release indexed | |
Features, dependencies and evidence for SGLang 0.5.19 | ||||
| TensorRT-LLM | 1.2.1Newest indexed | Architecture listed; exact checkpoint unknown | No supported release indexed | |
Features, dependencies and evidence for TensorRT-LLM 1.2.1 | ||||