Across 1 comparable benchmark, Kimi K3 has better reported results on 1 and worse on 0.
Limited evidence1 of 1 shared result countedHow we compare
Mistral Medium 3.1 is no longer available through its official API. API prices for this version are not shown.
Price basis: International rates vs listed rates. The price section below shows the applicable periods.
| Fact | Kimi K3 | Mistral Medium 3.1 |
|---|---|---|
| Context window | 1M tokens | 128K tokens |
| Input / 1M | $3.00 | β |
| Output / 1M | $15.00 | β |
| Blended (4:1) | $3.46 | β |
| Status | Available | Retired |
| Released | July 16, 2026 | August 12, 2025 |
Blended is 4:1 input to output with 90% cache hits at listed rates β set your own mix.
No shared benchmark here has a fixed 0β100 scale, so only the counted results are shown. How averages are built
Kimi K3 ahead on 1 of 1 counted
No fixed-scale average β see the rows in Benchmark results
No shared comparable results for Coding, Abstract reasoning, Mathematics, Science & health, Computer & tool use, Research & search, Long context, Multimodal, Cybersecurity.
Published scores, with matching modes where available. Gaps show Kimi K3 relative to Mistral Medium 3.1. Reading the results
| Benchmark | Kimi K3 | Mistral Medium 3.1 | Gap |
|---|---|---|---|
| Creative WritingWriting | 2083Normal Β· no tools | 1476Normal Β· no tools | 1.41Γ#6 vs #64 of 121 |
Writing
A dash means no result is available.
Time-of-day providers use peak-hour rates; fixed-price providers use standard rates. Off-peak discounts remain in the published details.
Drag the handle above the bar for cache hits, the one below it for the input : output mix Β· 5% steps
Blended price = effective input rate at 90% cache hits Γ 80% + output rate Γ 20%
A blended cost needs a published input and output price on both sides, and one of these two does not have them yet.
Kimi K3: International pricing. Moonshot official pricing verified 2026-10-21 (platform.kimi.ai/docs/pricing/chat)
Mistral Medium 3.1: Listed text rates; pricing tier not confirmed.
| Price / 1M tokens | Kimi K3 | Mistral Medium 3.1 | Ratio |
|---|---|---|---|
| Input | $3.00 | β | β |
| Output | $15.00 | β | β |
| Cached input | $0.30 | β | β |
| Cache write | β | β | β |
| Blended | $3.46 | β | β |
Estimates use the selected token mix. Cache writes, tool fees and long-context surcharges are excluded.
Reported cost per task, paired with the score from the same run. Settings and sources may differ; costs do not affect the verdict.
| Benchmark | Kimi K3 | Mistral Medium 3.1 |
|---|---|---|
| AA Intelligence Index v5.0.1 | Not reported | |
| AA-LCR | Not reported | |
| Agent Arena | $0.852 / task Median Β· Net improvement index (percentage points vs. average model): 0.51 Max (With Tools + Internet) | Not reported |
| Agent Arena Index | $0.9157 / task Median Β· Agent Arena Index rating (index points): 1093 Max (With Tools + Internet) | Not reported |
| Agents' Last Exam v1.0 | Not reported | |
| AIME 2026 | Not reported | |
| ARC-AGI-1 | $0.4798 / task Average Β· Score (%): 86.7 Thinking High (With Tools) $0.1806 / task Average Β· Score (%): 65.7 Thinking Low (With Tools) | Not reported |
| ARC-AGI-2 | $0.9472 / task Average Β· ARC-AGI-2 Score: 55.0 Thinking High (With Tools) $0.2541 / task Average Β· ARC-AGI-2 Score: 12.4 Thinking Low (With Tools) | Not reported |
| AutomationBench | $0.43 / task Average Β· Pass Rate: 22.7 Reported best (effort unspecified) Β· zapier-private-heldout | Not reported |
| CWE-bench v0 | Not reported | |
| DeepSWE | Not reported | |
| FrontierCode 2.0 | $4.475 / task Average Β· Weighted rubric score: 17.7 Thinking High (With Tools + Internet) $2.977 / task Average Β· Weighted rubric score: 11.1 Thinking Low (With Tools + Internet) | Not reported |
| FrontierCS 2 | Not reported | |
| FrontierSWE v2 | Not reported | |
| GDPx | $28.38 / task Average Β· GDPx β Score: 3.7 Highest (source-reported) (With Tools) Β· overall/score | Not reported |
| LiveBench 2.0 | $1.1961 / task Average Β· Overall score (equal-weight mean of 7 suite scores, 0-100): 63.4 Max (task-specific tools) | Not reported |
| SWE-sweep | Not reported | |
| Terminal-Bench 4.0 | $14.0029 / task Average Β· Resolution rate (%): 14.1 Max (With Tools) Β· vals-effort-study $13.4419 / task Average Β· Resolution rate (%): 24.2 Thinking High (With Tools) Β· vals-effort-study $12.6005 / task Average Β· Resolution rate (%): 7.1 Thinking Low (With Tools) Β· vals-effort-study 1 more settings$12.48 / task Average Β· Resolution rate (%): 17.2 Unspecified Reasoning (With Tools) Β· vals-mini-swe | Not reported |
| Terminal-Bench 4.1 | Not reported | |
| Vals Index v2.1 | $7.262 / task Average Β· GDP-weighted accuracy (%): 50.7 Reported best (effort unspecified) | Not reported |
| Vals Index v2.2 | Not reported | |
| Vibe Code Bench 1-100 | Not reported | |
| WeirdML v3 | $19.8848 / task Average Β· Official score (0-1; higher is better): 0.075 Thinking High (With Tools) | Not reported |
| Attribute | Kimi K3 | Mistral Medium 3.1 |
|---|---|---|
| Availability | Available | Retired |
| Provider | Moonshot AI | MistralAI |
| Model type | Reasoning model | Chat model |
| First release | July 16, 2026 | August 12, 2025 |
| Context window | 1M tokens | 128K tokens |
| Max output | 1M tokens | β |
| Total parameters | 2.8T | β |
| Active parameters | 104B | β |
| Architecture | Mixture of Experts (MoE) | Undisclosed |
| Documented capabilities | Reasoning, Multilingual, Coding, Function calling, Tool use, Structured output, Computer use, Web search, Long context | β |
| Input modalities | Text, Image, Video | Text |
| Output modalities | Text | Text |
| Weights | Open weights | Closed / API only |
| Licence | Kimi K3 License | β |
| Commercial use | Commercial use allowed, with conditions | Proprietary |
These are reported-score comparisons, not controlled head-to-head tests.
What people found comparing these models, or using either one. Comments also appear on the models' own pages.
Used Kimi K3 or Mistral Medium 3.1? Sign in with Google to share what worked, what did not, and which API or setup you tested.
No community comments on either model yet.
Alternatives with the most benchmark overlap for this matchup.