Kimi K3
Moonshot's agentic coding update to Kimi K2, positioned around long-horizon tool use, with open weights and publisher-reported coding and science benchmarks.
Routable on MicroRouter
Use moonshotai/kimi-k3 via the API →- Context
- 256K
- Per the announcement's spec table
- Price
- $0.60 / $2.40
- Per 1M input / output tokens; cached input at $0.15 per 1M
- Weights
- Released
- Modified MIT on Hugging Face, following the K2 licensing pattern
- Reasoning effort
- low / high
- Two publisher-described effort levels; high is the default
Publisher pricing
Pricing below comes from the release announcement and does not represent a MicroRouter route or price.
| Phase | Input / 1M | Output / 1M | Blended / 1M | Effective | Source |
|---|---|---|---|---|---|
| Standard | $0.60 | $2.40 | $1.05 | Aug 8, 2026 – open-ended | Moonshot AI ↗ |
Publisher-reported benchmarks
Results are grouped by unit. Percentage, Elo, and task-count results never share a scale.
| Benchmark | Reported result | Reported by | As of | Source |
|---|---|---|---|---|
| SWE-Bench Pro | 72.4% | Moonshot AI | Aug 8, 2026 | Announcement ↗ |
| Terminal Bench 2.1 | 84.1% | Moonshot AI | Aug 8, 2026 | Announcement ↗ |
| GPQA Diamond | 88.9% | Moonshot AI | Aug 8, 2026 | Announcement ↗ |
| DeepSWE v1.1 | 63.8% | Moonshot AI | Aug 8, 2026 | Announcement ↗ |
- — Moonshot reports the benchmark figures; MicroRouter has not reproduced them.
Independent measurements
Artificial Analysis runs its own evaluations and timing against each model's first-party API. These are the only numbers on this page not reported by the publisher.
Kimi K3 (max) ranks #6 of 448 model families on their Intelligence Index. Their headline uses the highest-scoring effort setting; every listed setting is below.
- Intelligence Index
- 59.7
- Composite of their evaluations
- Coding Index
- 76.2
- Coding evaluations only
- Blended price
- $6
- USD per 1M tokens, 3:1 input to output, first-party list
- Output speed
- 39
- Median tokens per second
- First answer token
- 54s
- Median seconds, including reasoning
| Setting | Intelligence Index | Coding Index | Blended price | Output speed | First answer token |
|---|---|---|---|---|---|
| Kimi K3 (max) ↗ | 59.7 | 76.2 | $6 | 39 | 54s |
| Kimi K3 (low) ↗ | 48.3 | 72.0 | $6 | 35 | 60s |
Source: Artificial Analysis ↗, read Sep 2, 2026. Measured on the model's first-party API, not a MicroRouter route.
What is not independently confirmed
- Every figure on this page is self-reported by the cited publisher. MicroRouter did not run these benchmarks.
- Benchmark harnesses are not standardised. Results sharing a name across vendors have not been confirmed to use identical runs.
- Publisher pricing can change, and introductory phases have stated end dates.
- Kimi K3 is routable through MicroRouter today; live pricing on its model page is authoritative over the publisher figures here.
Related coverage
Other release dossiers and the current MicroRouter catalog.
Sources
- Read at Moonshot AI ↗
Moonshot AI announcement
Published Aug 8, 2026
- Read at Artificial Analysis ↗
Artificial Analysis model page
Published Sep 2, 2026
Editorial reference revised 2026-09-02. Catalog status comes from the generated routing snapshot.