Solar System
loads high
Tech. quality95
UX94
Time3 min 44 s
Cost$0.28
Compare →
Mistral Medium 3.5
Loads without errors5/5#1 of 26
Tech. quality93/100#16 of 26
UX93/100#1 of 26
Total cost$1.49
Average time4 min 09 s#14 of 26
Total tokens199,885
API price: $1.5 / $7.5 per 1M input / output tokens · as of 17/10/2026 source ↗
What is Mistral Medium 3.5 good for?
Mistral's multimodal model optimised for agentic and coding use cases, with open weights.
Description based on the maker · source ↗
- Maker
- Mistral
- Released
- 28/04/2026
- Context window
- 256K tokens
- Model weights
- open (Modified MIT licence)
- Reasoning levels
- none high
How did Mistral Medium 3.5 do in the AI Lab tests?
- 5 of 5 outputs load without errors.
- Its best technical result was on the “Monitor test” prompt (97/100).
- In the objective perft chess test it passed 0 of 6 positions.
- Average time per prompt 4 min 09 s#14 of 26
- Average cost per prompt $0.30#18 of 26
- Technical quality 93/100#16 of 26
- Automatic UX rating 93/100#1 of 26
Outputs
All outputs of Mistral Medium 3.5 on the 5 prompts every other model got as well.
GTA-style game
loads high
Tech. quality89
UX100
Time4 min 09 s
Cost$0.30
Compare →
Monitor test
loads high
Tech. quality97
UX100
Time2 min 45 s
Cost$0.21
Compare →
New York & Central Park map
loads high
Tech. quality96
UX84
Time5 min 32 s
Cost$0.38
Compare →
Chess with an engine
loads high
Tech. quality89
UX87
Time4 min 36 s
Cost$0.32
Compare →
Cost = tokens priced with the API price list valid on the date shown next to the cost (prices change). ≈ = estimate (only the total token count is known), no ≈ = exact from input / output tokens. “Loads” = the page opens without errors; technical quality also checks behaviour during use – judge the quality in the viewer.