Qwen3.8 Max Preview vs Kimi K3 vs Kimi K2.7 Code Highspeed
Qwen3.8 Max Preview comes out ahead, 47 to 44 and 43 on our weighted score, and it is the cheaper option too.
- Our pick
Alibaba (Qwen)Qwen3.8 Max Preview
47/100- ECI—
- Price$2.00 / $6.00
- Context1M
Moonshot AIKimi K3
43/100- ECI157.6
- Price$3.00 / $15.00
- Context1.05M
- Moonshot AI
Kimi K2.7 Code Highspeed
44/100- ECI—
- Price$1.90 / $8.00
- Context262K
Qwen3.8 Max Preview is our pick
Qwen3.8 Max Preview is the better all-round choice, scoring 47/100 against Kimi K2.7 Code Highspeed (44) and Kimi K3 (43). It leads on price. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
- CapabilityNot enough dataNo independent benchmark covers every model here yet
- Lowest priceQwen3.8 Max PreviewQwen3.8 Max Preview $3.00 · Kimi K2.7 Code Highspeed $3.42 · Kimi K3 $6.00 per 1M tokens (3:1 blend)
- Longest contextKimi K3Kimi K3 1,048,576 · Qwen3.8 Max Preview 1,000,000 · Kimi K2.7 Code Highspeed 262,144 tokens
- Widest inputsSame inputsQwen3.8 Max Preview: Text, Images, Video · Kimi K3: Text, Images, Video · Kimi K2.7 Code Highspeed: Text, Images, Video
- Self-hostingKimi K3 and Kimi K2.7 Code HighspeedPublishes downloadable weights
| Measure | Weight | Qwen3.8 Max Preview | Kimi K3 | Kimi K2.7 Code Highspeed |
|---|---|---|---|---|
| Price | 50% | 27 | 13 | 25 |
| Inputs & features | 30% | 70 | 80 | 80 |
| Context window | 20% | 60 | 61 | 37 |
| Overall | 100% | 47/100 | 43/100 | 44/100 |
Left out because at least one model lacks the data: capability. The remaining weights were rescaled.
Every spec in one table
Highlighted cells lead their row. Dashes mean the data is not published.
| Specification | Qwen3.8 Max PreviewAlibaba (Qwen) | Kimi K3Moonshot AI | Kimi K2.7 Code HighspeedMoonshot AI |
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | — | 157.6 | — |
| ECI rank | — | #13 of 148 | — |
| GPQA DiamondGraduate-level science questions | — | 93.1% | — |
| FrontierMath Tiers 1–3Research-level mathematics | — | 72.2% | — |
| OTIS Mock AIME 2024–2025Competition mathematics | — | 97.2% | — |
| SimpleQA VerifiedShort factual questions | — | 50.6% | — |
| Price per million tokens | |||
| Input | $2.00 | $3.00 | $1.90 (best) |
| Output | $6.00 (best) | $15.00 | $8.00 |
| Cached input | — | $0.30 | — |
| Blended (3:1) | $3.00 (best) | $6.00 | $3.42 |
| Long-context rate | Same rate | Same rate | Same rate |
| Price source | Median of 6 providers | Official Moonshot AI API | Median of 11 providers |
| Limits | |||
| Context window | 1,000,000 tokens | 1,048,576 tokens (best) | 262,144 tokens |
| Max output | 131,072 tokens | 131,072 tokens | 262,144 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | Yes | Yes |
| PDFs | No | No | No |
| Audio | No | No | No |
| Video | Yes | Yes | Yes |
| Reasoning | Yes | Yeslow · high · max | Yes |
| Tool calling | Yes | Yes | Yes |
| Structured output | No | Yes | Yes |
| Availability | |||
| Weights | Proprietary | Open | Open |
| API model ID | — | kimi-k3 | — |
| API providers | 6 | 68 (best) | 11 |
| Released | Jul 19, 2026 | Jul 16, 2026 | Jun 12, 2026 |
| Knowledge cutoff | — | — | Jan 2025 |
What would a month cost?
Enter your expected volume in millions of tokens. List prices only; caching and batch discounts would lower these.
- Qwen3.8 Max Preview$32.00
- Kimi K3$60.00
- Kimi K2.7 Code Highspeed$35.00
Which should you choose?
Which is better: Qwen3.8 Max Preview, Kimi K3 or Kimi K2.7 Code Highspeed?
Qwen3.8 Max Preview is the better all-round choice, scoring 47/100 against Kimi K2.7 Code Highspeed (44) and Kimi K3 (43). It leads on price. The score weighs price 50%, inputs & features 30%, context window 20%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, Qwen3.8 Max Preview, Kimi K3 or Kimi K2.7 Code Highspeed?
Qwen3.8 Max Preview is cheaper at $2.00 input / $6.00 output per million tokens (median across 6 API providers). Kimi K2.7 Code Highspeed costs $1.90 input / $8.00 output per million tokens (median across 11 API providers); Kimi K3 costs $3.00 input / $15.00 output per million tokens (official Moonshot AI API price). At a typical mix of three input tokens to one output token, that is $3.00 per million tokens for Qwen3.8 Max Preview versus $3.42 for Kimi K2.7 Code Highspeed (1.1× as much) and $6.00 for Kimi K3 (2× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Qwen3.8 Max Preview has not been scored yet, Kimi K3 has an ECI of 157.6 and Kimi K2.7 Code Highspeed has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Qwen3.8 Max Preview, Kimi K3 and Kimi K2.7 Code Highspeed yet, so there is no like-for-like coding score. Test both on a sample of your own tasks. All three support tool calling for agent workflows.
Which has the bigger context window?
Kimi K3 has the largest context window at 1,048,576 tokens, against 1,000,000 for Qwen3.8 Max Preview and 262,144 for Kimi K2.7 Code Highspeed. Maximum output per response: Qwen3.8 Max Preview up to 131,072, Kimi K3 up to 131,072, Kimi K2.7 Code Highspeed up to 262,144 tokens.
Which can read images, PDFs, audio or video?
Qwen3.8 Max Preview accepts text, images and video; Kimi K3 accepts text, images and video; Kimi K2.7 Code Highspeed accepts text, images and video. They handle the same number of input types.
Are any of these open source?
Kimi K3 and Kimi K2.7 Code Highspeed publishes its weights and can be self-hosted; Qwen3.8 Max Preview is proprietary.
Which is newer?
Qwen3.8 Max Preview is the newest, released Jul 19, 2026. Kimi K3 came out Jul 16, 2026; Kimi K2.7 Code Highspeed came out Jun 12, 2026. Knowledge cutoff: Kimi K2.7 Code Highspeed Jan 2025.
How do you decide the winner?
Each model gets a 0–100 score on capability (50%, independent benchmark results); price (25%, blended price per million tokens (3 input : 1 output), log scale); inputs & features (15%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (10%, maximum tokens per request, log scale). Dimensions missing for any model are dropped and the remaining weights rescaled, so every model is judged on the same evidence. Specs and prices come from public model listings and the labs’ own API pages; capability scores come from independent benchmark runs. Data updated Oct 4, 2026.
