Model Follower.

Up to date

Rebuilt
2026-10-25 06:54 UTC
Release radar
117 on watch
Hub snapshot
2026-10-25
Re-checked
daily · every number sourced and dated

Models › Mistral-v0.2 family

Mistral-7B-Instruct-v0.2

Hugging Face ↗ Follow changes

mistralai Apache 2.0 out 3 years ago

7.24B parameters, text in, text out, 32K tokens of context.

Verified 2026-10-25 Rank #71 in LLMs · score 104.3 method v1 Sources 3 Downloads 30d 1.8M · #68 of 619 tracked

Can I run it?4.3GB at INT4Estimatefits a 16 GB cardmethodCommunity: Mistral-7B-Instruct-v0.2: architecture, parameters and context length | Frontier ML →
Can I use it?Apache 2.0Permissive; patent grantlicence file2026-10-24
Do people use it?1.8Mdownloads, 30 days#68 of 619 we track · top 11%+5 likes, downloads +7% this week3,270 likesHugging Face2026-10-24

Worth reading

Picked by rule from 6 tracked pieces: the lab's own post, the most shared, a field report, the latest.

6 best of 6 tracked

The 1 pick above, plus these 5. Top is the radar's score: discussion, source authority and the lab's own posts, fading with age.

  1. ArticleMistral-7B-Instruct-v0.2: architecture, parameters and context length | Frontier MLfrontiermlnp.com · undated
  2. ArticleMistral-7B-Instruct-v0.2 by Mistral AI_: Open-Weight Modelsavrn.com · undated
  3. ArticleMistral 7B Instruct v0.2compl-ai.org · undated
  4. ArticleMistral 7B Instruct V0.2 by mistralaillm.extractum.io · undated
  5. ArticleMistral-7B-Instruct v0.2 Overviewemergentmind.com · undated

Where it's used

Hugging Face does not say who downloads a model or with what. These are the proxies: API providers listing it, quantized builds, fine-tunes/adapters/merges, and Spaces that load it.

No API provider serves it, so its downloads are people running it themselves: 112 quantized builds and 100 Spaces.

  • As a service0no API provider listed
  • On own hardware112quantized builds
  • Built on2,7181,157 fine-tunes · 1,208 adapters · 353 merges
  • In apps100+Hugging Face Spaces

Top Spaces running it: KingNish/OpenGPT-4o ♥ 4,302 · KingNish/JARVIS ♥ 292 · allenai/WildBench ♥ 232 · JournalistsonHF/ai-scraper ♥ 150

Will it fit?

Weights only, an estimate: leave room for the KV cache. Method

Fits a 16 GB card at INT4

About 4.3 GB at INT4, 8.7 GB at FP8, 17 GB at BF16: weights only, so leave room for the KV cache.

16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 4.3 GB of weights (estimate; KV cache and activations extra)INT44.3 GBFP8: about 8.7 GB of weights (estimate; KV cache and activations extra)FP88.7 GBBF16: about 17 GB of weights (estimate; KV cache and activations extra)BF1617 GB
16 GB card64 GB Mac2×80 GB8×80 GBINT4: about 4.3 GB of weights (estimate; KV cache and activations extra)INT44.3 GBFP8: about 8.7 GB of weights (estimate; KV cache and activations extra)FP88.7 GBBF16: about 17 GB of weights (estimate; KV cache and activations extra)BF1617 GB

Estimate params × bytes per param × 1.2 · how we estimate

The story so far

Weights, every dated article, and repo commits on one rail.

22 repo commits since the weights landed

Teal is the weights (2023-12-11), grey ticks are repo commits.

JanFebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOct2025-07-24: 1 repo commit2025-07-23: 1 repo commit2024-09-27: 1 repo commit2024-08-02: 1 repo commit2024-07-30: 1 repo commit2024-07-24: 1 repo commit2024-07-03: 1 repo commit2024-06-24: 1 repo commit2024-06-20: 1 repo commit2024-03-24: 1 repo commit2024-02-29: 1 repo commit2023-12-15: 2 repo commits2023-12-14: 2 repo commits2023-12-13: 1 repo commit2023-12-12: 1 repo commit2023-12-11: 5 repo commitsWeights 2023-12-11 · 2023-12-11Weights 2023-12-11now
JanFebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOctNovDecJanFebMarAprMayJunJulAugSepOct2025-07-24: 1 repo commit2025-07-23: 1 repo commit2024-09-27: 1 repo commit2024-08-02: 1 repo commit2024-07-30: 1 repo commit2024-07-24: 1 repo commit2024-07-03: 1 repo commit2024-06-24: 1 repo commit2024-06-20: 1 repo commit2024-03-24: 1 repo commit2024-02-29: 1 repo commit2023-12-15: 2 repo commits2023-12-14: 2 repo commits2023-12-13: 1 repo commit2023-12-12: 1 repo commit2023-12-11: 5 repo commitsWeights 2023-12-11 · 2023-12-11Weights 2023-12-11now

What people built on it

2,830 derivatives on the Hub. Which build to run, by runtime: the most downloaded in each class.

2,830 derivatives on the Hub, led by 1,208 adapters

From the Hugging Face model tree; quantized builds in teal, fine-tunes, merges and adapters in greys.

112 quantized (Hugging Face model tree)1,157 fine-tunes (Hugging Face model tree)353 merges (Hugging Face model tree)1,208 adapters (Hugging Face model tree)112 quantized1,157 fine-tunes353 merges1,208 adapters
112 quantized (Hugging Face model tree)1,157 fine-tunes (Hugging Face model tree)353 merges (Hugging Face model tree)1,208 adapters (Hugging Face model tree)112 quantized1,157 fine-tunes353 merges1,208 adapters
GGUF TheBloke/Mistral-7B-Instruct-v0.2-GGUF for llama.cpp, Ollama, LM Studio 3M downloads AWQ TheBloke/Mistral-7B-Instruct-v0.2-AWQ for vLLM, SGLang 3.2M downloads

All variants →

Reference

Everything else, folded: the full tables, every source.

What the card says

1 cited sentence

Summary generated, cited

This instruct fine-tuned model is released under the Apache 2.0 license.[1][2]

  1. license
  2. “is an instruct fine-tuned version”

Generated 2026-10-07 15:00 UTC by parasail-gemma-4-26b-a4b-it; each numeral points at the fact it rests on. Not a verdict.

Use-case evidence

2 use cases

What it is good at 2 use cases

Labels are evidence, not verdicts: measured a third-party leaderboard row for this model; author-reported a number the model author published in the card or eval files; structural a deterministic signal from the name, config or chat template, or a decision-model estimate. Each chip links to its source. All use cases.

Facts

20 fields

Listed in LLMs

Facts

Pipeline tagtext-generation
Librarytransformers
ArchitecturesMistralForCausalLM
Model typemistral
Dtypebfloat16
Params by dtypeBF16: 7,241,732,096
Size bandM
MoEno
Context length32,768 (config.json max_position_embeddings)
PapersarXiv:2310.06825
LicenseApache 2.0 (apache-2.0)
Hub snapshot2026-10-25

Servability

vLLM · SGLang

Servability

vLLM: supported · MistralForCausalLM · 7050ef881ff0835f6890a47db1859a24c65481d6SGLang: supported · MistralForCausalLM · 773cb938fe9484a8f7e7e46fb1b1fe96412c59ed

vLLM registry (7050ef881ff0835f6890a47db1859a24c65481d6) · SGLang registry (773cb938fe9484a8f7e7e46fb1b1fe96412c59ed) · Matched by architecture class name against each project's registry on main; 'supported' means the architecture is registered, not that this checkpoint was tested.

Hardware fit

3 estimates

Hardware fit ESTIMATE

Weights-only memory for Mistral-7B-Instruct-v0.2: params × bytes/param × 1.2 overhead. KV cache and activations are extra. How this is computed.

BF16 (2 bytes/param)17 GB · Single GPU (<24 GB)
FP8 (1 byte/param)8.7 GB · Single GPU (<24 GB)
INT4 (0.5 bytes/param)4.3 GB · Single GPU (<24 GB)

Weights only (params × bytes/param × 1.2 overhead); KV cache and activations extra.

All variants

40 variants

Lineage

Original model Family: mistralai/Mistral-v0.2 2,830 derivatives on the Hub

quantized: 112finetune: 1,157merge: 353adapter: 1,208

Top variants (40 shown, ranked by downloads)

Relation
Quant
Size
40 rows
Variant Relation Quant Params Band License Author Likes Downloads
TheBloke/Mistral-7B-Instruct-v0.2-AWQ quantized AWQ 7.24B M apache-2.0 TheBloke 52 3.2M
TheBloke/Mistral-7B-Instruct-v0.2-GGUF quantized GGUF — — apache-2.0 TheBloke 522 3M
TheBloke/Mistral-7B-Instruct-v0.2-GPTQ quantized GPTQ 7.24B M apache-2.0 TheBloke 56 2.9M
abacusai/bigstral-12b-32k finetune — 12.5B L apache-2.0 abacusai 44 199.6k
mozilla-ai/Mistral-7B-Instruct-v0.2-llamafile quantized GGUF — — apache-2.0 mozilla-ai 26 175.2k
mlc-ai/Mistral-7B-Instruct-v0.2-q4f16_1-MLC quantized — — — — mlc-ai 4 82.7k
X-iZhang/libra-llava-med-v1.5-mistral-7b merge — 7.57B M apache-2.0 X-iZhang 2 70.8k
Kabster/Bio-Mistralv2-Squared merge — 7.24B M apache-2.0 Kabster 1 48.1k
QuantFactory/Lily-Cybersecurity-7B-v0.2-GGUF quantized GGUF — — apache-2.0 QuantFactory 28 46.3k
NeuralNovel/Mini-Mixtral-v0.2 merge — 12.9B L apache-2.0 NeuralNovel 4 39.3k
Mihaiii/Metis-0.3 finetune — 7.24B M apache-2.0 Mihaiii 1 37.2k
segolilylabs/Lily-Cybersecurity-7B-v0.2 finetune — 7.24B M apache-2.0 segolilylabs 164 27.7k
Edentns/DataVortexM-7B-Instruct-v0.1 finetune — 7.24B M cc-by-nc-sa-4.0 Edentns 0 35k
refarde/Mistral-7B-Instruct-v0.2-Ko-S-Core finetune — 7.24B M apache-2.0 refarde 1 34.1k
NeuralNovel/Panda-7B-v0.1 finetune — 7.24B M apache-2.0 NeuralNovel 14 25.4k
OpenVINO/Mistral-7B-Instruct-v0.2-int4-ov quantized int4 — — apache-2.0 OpenVINO 1 24.8k
NeuralNovel/Gecko-7B-v0.1 finetune — 7.24B M apache-2.0 NeuralNovel 6 23.4k
second-state/Mistral-7B-Instruct-v0.2-GGUF quantized GGUF — — apache-2.0 second-state 1 20.8k
EmbeddedLLM/Mistral-7B-Merge-14-v0.3 merge — 7.24B M apache-2.0 EmbeddedLLM 8 19.2k
EmbeddedLLM/Mistral-7B-Merge-14-v0.3-ft-step-15936 merge — 7.24B M apache-2.0 EmbeddedLLM 0 17k
EmbeddedLLM/Mistral-7B-Merge-14-v0.3-ft-step-9984 merge — 7.24B M apache-2.0 EmbeddedLLM 0 17k
Cartinoe5930/DARE-Merging merge — 7.24B M apache-2.0 Cartinoe5930 1 14.1k
mradermacher/Mistral-7B-Instruct-v0.2-i1-GGUF quantized GGUF — — apache-2.0 mradermacher 1 14k
Praneeth/StarMix-7B-slerp merge — 7.24B M apache-2.0 Praneeth 2 13.7k
NeuralNovel/Mistral-7B-Instruct-v0.2-Neural-Story finetune — 7.24B M apache-2.0 NeuralNovel 29 12.3k
QuantFactory/Mental-Health-FineTuned-Mistral-7B-Instruct-v0.2-GGUF quantized GGUF — — apache-2.0 QuantFactory 8 13.2k
bartowski/Aethora-7b-V1-e-1-GGUF quantized GGUF — — — bartowski 1 13.4k
s3nh/nsfw-noromaid-mistral-instruct merge — 7.24B M apache-2.0 s3nh 2 12.8k
GRMenon/mental-health-mistral-7b-instructv0.2-finetuned-V2 adapter — — — apache-2.0 GRMenon 40 10.7k
osanseviero/mistral-instruct-frankenmerge merge — 8.99B M apache-2.0 osanseviero 2 12.3k
osanseviero/mistral-instruct-moe-experimental merge — 12.9B L apache-2.0 osanseviero 2 12.3k
paulilioaica/Hugo-7B-slerp merge — 7.24B M apache-2.0 paulilioaica 0 12.3k
Radu1999/Mistral-Instruct-Ukrainian-slerp merge — 7.24B M apache-2.0 Radu1999 1 12.1k
osanseviero/mistral-instruct-slerp merge — 7.24B M apache-2.0 osanseviero 1 12.1k
nasiruddin15/Mistral-grok-instract-2-7B-slerp merge GGUF 7.24B M mit nasiruddin15 2 10.5k
Yongxin-Guo/trace-uni finetune — 7.55B M apache-2.0 Yongxin-Guo 5 10.3k
NeuralNovel/Tanuki-7B-v0.1 finetune — 7.24B M apache-2.0 NeuralNovel 8 10k
cognitivetech/Mistral-7b-Inst-0.2-Bulleted-Notes_GGUF quantized GGUF — — apache-2.0 cognitivetech 3 10k
Eric111/Mistral-7B-Instruct_v0.2_UNA-TheBeagle-7b-v1 merge — 7.24B M cc-by-nc-nd-4.0 Eric111 1 9.9k
MaziyarPanahi/Mistral-7B-Instruct-v0.2-GGUF quantized GGUF — — apache-2.0 MaziyarPanahi 1 9.5k

Repository changelog

22 commits

Repository changelog 22 commits

DateCommitSHA
2025-07-24 Update README.md 63a8b08
2025-07-23 Update README.md abb3692
2024-09-27 Expose metadata link to next version of the model (#161) 3ad372f
2024-08-02 Update README.md 36d7e54
2024-07-30 Remove warning from README.md (#149) 83b95e8
2024-07-24 Update README.md 05bed4c
2024-07-03 Align tokenizer with mistral-common (#141) 250544c
2024-06-24 [AUTO] CVST Tokenizer Badger (#140) 9925900
2024-06-20 Updating chat template, trying to make it match with `mistral_common` (#136) 1296dc8
2024-03-24 Update README.md (#70) 41b61a3
2024-02-29 Enable inference & update examples in README to be compatible with soon-to-come ChatWidget (#54) cf47bb3
2023-12-15 Update config.json b70aa86
2023-12-15 Update README.md 98d9793
2023-12-14 Update config.json cb35ee2
2023-12-14 Readme from_pretrained model version fix (#18) a8d85f7
2023-12-13 Inputs for widget using chat format (#6) 235b80c
2023-12-12 Adding `safetensors` variant of this model (#1) 5a1029f
2023-12-11 Update README.md c72e5d1
2023-12-11 Update README.md 76399c6
2023-12-11 Create README.md f1586aa
2023-12-11 Add v0.2 dca6e4b
2023-12-11 initial commit e68448a

Articles and coverage

6 pieces

Ranked by source signals (HN, Lobsters, authority domains, official posts) and recency. Summaries are generated from a stored copy of the article text; every sentence cites a span of that text, and nothing is shown that failed the citation check.

Analysis6

Mistral-7B-Instruct-v0.2: architecture, parameters and context length | Frontier ML
frontiermlnp.com382 words
Mistral-7B-Instruct-v0.2 is a 7.0B parameter model where 78% of the weights are feed-forward and 19% are attention. At 1K tokens, BF16 weights require 14.6 GB, but serving a full 32K context demands a 24 GB RTX 4090 memory floor. The architecture features 32 transformer blocks with 4096-wide vectors and Grouped Query Attention collapsing 32 query heads onto 8 key/value heads.written, reviewed, cited
Mistral 7B Instruct v0.2: benchmarks — ModelCap #271
modelcap.ai834 words
Mistral 7B Instruct v0.2 is a language model from Mistral AI that ranks #338 of 348 on ModelCap's current board with an Index score of 7.2. The model first appeared in the catalogue on 2023-12-11 and has accumulated 1746964 Hugging Face 30d downloads while being released under an apache-2.0 license.written, reviewed, cited
  • The model explicitly lacks support for reasoning, tool use, structured output, and moderation capabilities.
  • Declared output limits are unknown because a model-wide maximum has not been independently verified.
  • The available weights are downloadable under an OSI or free-culture license, though this does not confirm full training stack openness.
Mistral-7B-Instruct-v0.2 by Mistral AI_: Open-Weight Model
savrn.com1,693 words
Mistral-7B-Instruct-v0.2 is an open-weight text generation model from Mistral AI released under Apache License 2.0. It features 7.2B parameters and a 32,768-token context window, requiring about 17.4 GB of GPU memory at 16-bit precision on a 1x MI300X setup costing $1.85 an hour. The model draws 1.7M downloads a month, scores 30.84 on MMLU-Pro, and has a 29.5 GB repo size with weights last updated 2025-07-24.written, reviewed, cited
  • The model card explicitly states that Mistral lists Mistral 7B Instruct v0.3 as retired as of 30 March 2025.
  • A third-party evaluation reports the MMLU-Pro score as 30.84, indicating a 2023 instruct model rather than a current reasoning model.
  • The repository duplicates weights in bin and safetensors formats, totaling 29.5 GB for the full release.
Mistral 7B Instruct v0.2
compl-ai.org4,748 words
The Mistral 7B Instruct v0.2 model is evaluated against technical requirements of the EU AI Act, focusing on robustness and cyberattack resilience. It achieves a 0.57 aggregate score on robustness benchmarks and a 0.24 aggregate score on cyberattack resilience tests. These results are compared to the Llama 2 70b Chat baseline, which scores 0.6 on robustness and 0.43 on cyberattack resilience.written, reviewed, cited
  • Corrigibility is currently not evaluated because it lacks a clear technical definition and measurable benchmarks.
  • The toxicity and bias of the training dataset are not scored for Mistral 7B Instruct v0.2.
Mistral 7B Instruct V0.2 by mistralai
llm.extractum.io
Mistral-7B-Instruct v0.2 Overview
emergentmind.com1,146 words

Papers

1 paper

Papers 1

arXiv links found in the model card and its tags. "Peer-reviewed" is shown only with evidence: a published venue in OpenAlex (measured) or the authors' own journal reference on arXiv (reported).

Similar models

3 models

Similar models

  • Llama-3.1-8B-Instruct · 8.03B
    same size band (M), same task (text-generation), more providers
  • Qwen2.5-7B-Instruct · 7.62B
    same size band (M), same task (text-generation), more providers, permissive license
  • Mistral-7B-v0.1 · 7.24B
    same size band (M), same task (text-generation), permissive license

Sources

3 sources

Sources

Hub snapshot: 2026-10-25 · HF API: 2026-10-24 21:33 UTC · Providers: 2026-10-24 21:44 UTC · Articles: 2026-10-24 21:45 UTC · Summary: 2026-10-07 15:00 UTC · Last verified 2026-10-25

Follow Mistral-7B-Instruct-v0.2

Watch this model

Email alerts are not open yet; the feeds above are the same data. See feeds for the item schema.