Grok 4.20 (Reasoning) vs Kimi K2.6 vs Step 5 Preview
No independent capability benchmark covers every model here yet. The score uses inputs & features and context window.
Price in rating: Excluded · Include price
xAI
Grok 4.20 (Reasoning)
74/100- ECI152.3
- Price$1.25 / $2.50
- Context1M
Moonshot AI
Kimi K2.6
65/100- ECI151.6
- Price$0.95 / $4.00
- Context262K
StepFun
Step 5 Preview
74/100- ECI—
- Price$1.00 / $2.70
- Context1M
Capability evidence is incomplete
There is not enough shared capability evidence to name an overall winner. The scores (Grok 4.20 (Reasoning) 74/100, Step 5 Preview 74/100, Kimi K2.6 65/100) compare the available specifications only. The score weighs inputs & features 60%, context window 40%. 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 priceStep 5 PreviewStep 5 Preview $1.43 · Grok 4.20 (Reasoning) $1.56 · Kimi K2.6 $1.71 per 1M tokens (3:1 blend)
- Longest contextGrok 4.20 (Reasoning) and Step 5 PreviewGrok 4.20 (Reasoning) 1,000,000 · Step 5 Preview 1,000,000 · Kimi K2.6 262,144 tokens
- Widest inputsSame inputsGrok 4.20 (Reasoning): Text, Images, PDFs · Kimi K2.6: Text, Images, Video · Step 5 Preview: Text, Images, Video
- Self-hostingKimi K2.6 and Step 5 PreviewPublishes downloadable weights
| Measure | Weight | Grok 4.20 (Reasoning) | Kimi K2.6 | Step 5 Preview |
|---|---|---|---|---|
| Inputs & features | 60% | 80 | 80 | 80 |
| Context window | 40% | 64 | 43 | 64 |
| Specs only | 100% | 74/100 | 65/100 | 74/100 |
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 | |||
|---|---|---|---|
| Capability | |||
| Capabilities Index (ECI) | 152.3 (best) | 151.6 | — |
| ECI rank | #59 of 158 (best) | #61 of 158 | — |
| GPQA DiamondGraduate-level science questions | 89.3% | 90.8% (best) | — |
| FrontierMath Tiers 1–3Research-level mathematics | 44.9% | 57.2% (best) | — |
| OTIS Mock AIME 2024–2025Competition mathematics | 92.2% | 96.1% (best) | — |
| SWE-bench VerifiedFixing real GitHub issues | — | 76.7% | — |
| SimpleQA VerifiedShort factual questions | 30.2% | 34.9% (best) | — |
| Price per million tokens | |||
| Input | $1.25 | $0.95 (best) | $1.00 |
| Output | $2.50 (best) | $4.00 | $2.70 |
| Cached input | $0.20 | $0.16 | $0.10 (best) |
| Blended (3:1) | $1.56 | $1.71 | $1.43 (best) |
| Long-context rate | Over 200K: $2.50 / $5.00 | Same rate | Same rate |
| Price source | Official xAI API | Official Moonshot AI API | Official StepFun (Global) API |
| Limits | |||
| Context window | 1,000,000 tokens (best) | 262,144 tokens | 1,000,000 tokens (best) |
| Max output | 30,000 tokens | 262,144 tokens | 1,000,000 tokens (best) |
| Inputs and features | |||
| Text | Yes | Yes | Yes |
| Images | Yes | Yes | Yes |
| PDFs | Yes | No | No |
| Audio | No | No | No |
| Video | No | Yes | Yes |
| Reasoning | Yes | Yes | Yeslow · medium · high |
| Tool calling | Yes | Yes | Yes |
| Structured output | Yes | Yes | Yes |
| Availability | |||
| Weights | Proprietary | Open | Open |
| API model ID | grok-4.20-0309-reasoning | kimi-k2.6 | step-5-preview |
| API providers | 14 | 64 (best) | 21 |
| Released | Mar 9, 2026 | Apr 20, 2026 | Sep 16, 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.
Grok 4.20 (Reasoning)$17.50
Kimi K2.6$17.50
Step 5 Preview$15.40
Which should you choose?
Which is better: Grok 4.20 (Reasoning), Kimi K2.6 or Step 5 Preview?
There is not enough shared capability evidence to name an overall winner. The scores (Grok 4.20 (Reasoning) 74/100, Step 5 Preview 74/100, Kimi K2.6 65/100) compare the available specifications only. The score weighs inputs & features 60%, context window 40%. None of the shared benchmarks cover every model here yet, so capability is left out of this verdict.
Which is cheaper, Grok 4.20 (Reasoning), Kimi K2.6 or Step 5 Preview?
Step 5 Preview is cheaper at $1.00 input / $2.70 output per million tokens (official StepFun (Global) API price). Grok 4.20 (Reasoning) costs $1.25 input / $2.50 output per million tokens (official xAI API price); Kimi K2.6 costs $0.95 input / $4.00 output per million tokens (official Moonshot AI API price). At a typical mix of three input tokens to one output token, that is $1.43 per million tokens for Step 5 Preview versus $1.56 for Grok 4.20 (Reasoning) (1.1× as much) and $1.71 for Kimi K2.6 (1.2× as much).
Which scores higher on benchmarks?
There is no independent benchmark that covers all three models yet. Grok 4.20 (Reasoning) has an ECI of 152.3, Kimi K2.6 has an ECI of 151.6 and Step 5 Preview has not been scored yet.
Which is better for coding?
There are no published SWE-bench Verified results for Grok 4.20 (Reasoning) and Step 5 Preview 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?
Grok 4.20 (Reasoning) and Step 5 Preview have the largest context windows (1,000,000 and 1,000,000 tokens), against 262,144 for Kimi K2.6. Maximum output per response: Grok 4.20 (Reasoning) up to 30,000, Kimi K2.6 up to 262,144, Step 5 Preview up to 1,000,000 tokens.
Which can read images, PDFs, audio or video?
Grok 4.20 (Reasoning) accepts text, images and PDFs; Kimi K2.6 accepts text, images and video; Step 5 Preview accepts text, images and video. They handle the same number of input types.
Are any of these open source?
Kimi K2.6 and Step 5 Preview publishes its weights and can be self-hosted; Grok 4.20 (Reasoning) is proprietary.
Which is newer?
Step 5 Preview is the newest, released Sep 16, 2026. Kimi K2.6 came out Apr 20, 2026; Grok 4.20 (Reasoning) came out Mar 9, 2026. Knowledge cutoff: Kimi K2.6 Jan 2025.
How do you decide the winner?
This rating uses inputs & features (60%, image, PDF, audio and video input, tool calling, structured output and reasoning); context window (40%, 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 Nov 18, 2026.