LOADING DATASHEET
LOADING DATASHEET
by Mistral
RANKED #175 OF 316 ASSISTANTS · OVERALL #249 OF 493 RANKED · VIBE SCORE 6.2 · 223 VOICES ·
COPY BADGE pastes a README snippet with the rank SVG.
Mistral Medium 3.5 is a dense 128B instruction-following model from Mistral AI. It supports text and image inputs with text output, and is designed for agentic workflows, coding, and complex...
42 mentions
40 mentions
26 weeks · 231 voices in this window
QUIETLY DEGRADING
Recent vibe and chatter are both sliding.
Open weights. You can use a hosted service, or download it and run it yourself, free.
Open weights: download Mistral Medium 3.5 once and it's yours. No subscription, no rate limits, works offline.
128B dense parameters, a Q4 file is about 75 GB
Custom PC with 3x Nvidia GeForce RTX 3090 24GB
72 GB VRAM PLUS 64 GB RAM
~$2,600 USEDIQ3_XXS or Q3_K quant with partial CPU offload, ~8 tok/s, 8k context
Apple Mac Studio M2 Ultra with 128 GB Unified Memory
128 GB UNIFIED MEMORY
~$3,400 USEDQ4_K_M quant in full memory, ~14 tok/s, 32k context
Custom workstation with 4x Nvidia RTX 6000 Ada 48GB
192 GB VRAM PLUS 256 GB RAM
~$28,000 NEWFP8 or FP16 full precision, ~32 tok/s, 128k context
The three cards above are researched picks. Search the machine you actually have — we only say yes if it should reply at a usable speed, not just load the weights and crawl.
HARDWARE PICKS & PRICES ARE RESEARCHED FROM THE LIVE WEB AND REFRESHED AUTOMATICALLY. TREAT THEM AS BALLPARK, NOT GOSPEL.
Everyday jobs on each of those rigs: how long Mistral Medium 3.5 takes, and how good the crowd says it is at that kind of work.
| THE JOB | BARE MINIMUMCustom PC with 3x Nvidia GeForce RTX 3090 24GB | SWEET SPOTApple Mac Studio M2 Ultra with 128 GB Unified Memory | FULL POWERCustom workstation with 4x Nvidia RTX 6000 Ada 48GB |
|---|---|---|---|
| Write an emaila paragraph or two, drafted from a one-line brief | ~63 S★★★★★ | ~36 S★★★★★ | ~16 S★★★★★ |
| Summarize a documenta long report boiled down to the points that matter | ~3.1 MIN★★★★★ | ~1.8 MIN★★★★★ | ~47 S★★★★★ |
| Build a websitea small landing page, markup and styles together | ~17 MIN★★★★★ | ~9.5 MIN★★★★★ | ~4.2 MIN★★★★★ |
| Build a backendan API with routes, storage and tests | ~52 MIN★★★★★ | ~30 MIN★★★★★ | ~13 MIN★★★★★ |
| Build a gamea playable browser game in one file | ~31 MIN★★★★★ | ~18 MIN★★★★★ | ~7.8 MIN★★★★★ |
STARS ARE THE COMMUNITY SCORE FOR THAT KIND OF WORK, DOCKED FOR HOW SQUASHED THE WEIGHTS GET AT EACH TIER. TIMES ARE COMPUTED FROM RESEARCHED SPEEDS. BALLPARK, NOT GOSPEL.
Not enough discussion about Mistral Medium 3.5 yet to call it. We found 223 posts but people did not say much either way.
142 POSTS · 81 COMMENTS · REDDIT · GITHUB · DEV.TO · LEMMY · HACKER NEWS · X
Checked picks first: tags say why you would switch, stars say how fully each one stands in for Mistral Medium 3.5.
Mistral Medium 3.5 is the only non-Chinese open-source model in the top 25 of the SWE-Bench Verified benchmark.
Given the amount of resources of Mistral, it seems like a miracle that they are still relevant
Mistral 3.5 out now!
Mistral medium 3.5 is out
Mistral Medium 3.5: A reliability first open source model from Europe
Mistral Medium 3.5: A reliability first open model from Europe
[New Model and More] Remote Agents in Vibe, Powered by Mistral Medium 3.5 in Public Preview
Mistral 3.5 Fixes
My Take on Mistral Medium 3.5 After a Month of Exclusive Use
Mistral Large 4 is a good model
Mistral Medium 3.5 128B is launched
Mistral medium 3.5 benchmarks