Solar System
broken default fixed version →
Overall0
Score0/8
Tech67
UX / UI0 / 0
Time2 min 58 s
Cost$0.019
Compare →
Codestral 25.08
MistralReleased 30/07/2025
Overall28/100#23 of 27
Tech79/100#25 of 27
UX / UI21 / 8#26 of 27
Score11.5/40
Total cost$0.075
Average time2 min 14 s#8 of 27
Total tokens84,769
API price: $0.3 / $0.9 per 1M input / output tokens · as of 17/10/2026 source ↗
What is Codestral 25.08 good for?
Mistral's specialised code model for low-latency, high-frequency tasks such as fill-in-the-middle and code generation.
Description based on the maker · source ↗
- Maker
- Mistral
- Released
- 30/07/2025
- Context window
- 128K tokens
How did Codestral 25.08 do in the AI Lab tests?
- 4 of 5 outputs load without errors.
- Its best technical result was on the “New York & Central Park map” prompt (100/100).
- Did not work (original output): Solar System.
- In the objective perft chess test it passed 0 of 6 positions.
- Average time per prompt 2 min 14 s#8 of 27
- Average cost per prompt $0.015#6 of 27
- Overall score 28/100#23 of 27
- Technical health 79/100#25 of 27
- Automatic UX rating 21/100#26 of 27
Outputs
All outputs of Codestral 25.08 on the 5 prompts every other model got as well.
GTA-style game
loads default
Overall38
Score4/8
Tech84
UX / UI0 / 0
Time2 min 07 s
Cost$0.015
Compare →
Monitor test
loads default
Overall26
Score2.5/8
Tech69
UX / UI0 / 0
Time1 min 12 s
Cost$0.00854
Compare →
New York & Central Park map
loads default
Overall44
Score3/8
Tech100
UX / UI78 / 0
Time1 min 14 s
Cost$0.00714
Compare →
Chess with an engine
loads default
Overall32
Score2/8
Tech74
UX / UI25 / 40
Time3 min 39 s
Cost$0.025
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 health also checks behaviour during use – judge the quality in the viewer.