DeepSWEmini-swe-agent
67%±2%
- bench
- DeepSWE
- version
- v1.1
- split
- public 113 tasks
- harness
- mini-swe-agent
- effort
- xhigh
- n
- 113
- metric
- Pass@1
- value
- 67%
- ci
- ±2%
- $/task
- $5.50
- tokens
- 71k out
- steps
- 87
grok-4.6Context 500k on the catalog, narrower than the 1M Claude / GPT-5.6 windows.
Ships across four xAI surfaces: the API, Grok Build, Cursor, and Grok Bot. Grok Bot never discloses its own default model and offers no picker, so a Bot run is not a Grok 4.6 run you can prove.
Use Grok 4.6 when you already pay xAI or want Continuum’s hosted Grok. DeepSWE official: 67% ±2% at xhigh, $5.50 per task, 71k out, 87 steps. Tight CI.
We did not open a vals.ai card for this identity. Chip omitted.
Ranked view: Best coding models: the independent leaderboard puts this row and every other card that carries an independent coding score on one board, with the confidence intervals left visible.
Copied from AA or vals HTML we opened. 0.0% placeholder bars are omitted. Do not average these into the hero tiles, and do not invent CI, $/task, or steps for Artificial Analysis.
| Bench | Printed | Note | Source | URL | As of |
|---|---|---|---|---|---|
| Official + OR list pair | $2 / $6 per 1M | Printed xAI embedded JSON and OpenRouter catalog. Cache read $0.50/M. ≥200k override $4 / $12. | xAI models docs | source | 2026-08-19 |
| AA task cost (their table) | about $0.84 per AA task | Printed AA cost footnote on the Grok 4.6 high tile. Not DeepSWE $5.50/task. | Artificial Analysis Grok 4.6 | source | 2026-08-19 |
batchDiscountPercent 0. Do not invent a batch discount.Plus is $25/mo with $25 weekly hosted usage. The Mac app stays free with your own keys. Get Plus is not Download for Mac.
xAI does not publish Grok 4.6 weights. No HF button.
batchDiscountPercent 0. Efforts low / medium / high / xhigh, default high.Lab card and OpenRouter listing only. No Hugging Face Files button, because there is no official repo to point at.
Fetched OpenRouter catalog on 2026-08-19. hugging_face_id was empty.
Continuum hosts grok-4.6 (19 Aug 2026). grok-4.5 left the hosted lineup on 17 Aug 2026.
We list first-party and OpenRouter, plus Continuum only when the live public allowlist named the id. This is not a 15-host routing table.
DeepSWE official: $5.50 per task at xhigh (71k out, 87 steps). That is the hero economics number.
Official xAI + OpenRouter, 19 Aug 2026: $2 / $6 per million, cache read $0.50 /M. Catalog web search $0.005 per call. ≥200k override $4 / $12. Batch discount printed as 0.
Artificial Analysis footnote, 19 Aug 2026, Grok 4.6 high: $2 / $6, about $0.84 per AA task. The AA tile is Intelligence Index 61. AA did not publish a DeepSWE-style CI or step count we will copy onto the chip.
| Claim | Source | As of |
|---|---|---|
| DeepSWE 67% ±2% at xhigh | DeepSWE official board | 2026-08-13 |
| xAI embedded JSON: $2 / $6 / cache $0.50; 2× at 200k; batchDiscountPercent 0; default effort high | xAI models docs | 2026-08-19 |
| OR catalog $2 / $6, cache $0.50, search $0.005, ≥200k $4 / $12 | OpenRouter catalog | 2026-08-19 |
| Hosted id grok-4.6 | GET /v1/chat/hosted/models/public | 2026-08-19 |
| OpenRouter id x-ai/grok-4.6, context 500,000 | OpenRouter /api/v1/models | 2026-08-19 |
Same official DeepSWE harness (mini-swe-agent, public 113 tasks), fetched 2026-08-19. Different effort labels are the lab's own setting on that board, shown here rather than normalized.
Continue through xAI's model family with Grok 4.5.
Lab posts pick a harness, an effort, a split, and sometimes a private eval set. DeepSWE publishes the official mini-swe-agent row with cost, tokens, and steps. Scale SWE-bench Pro only counts when the public shared-harness board has a row we can fetch. A higher lab number is usually a different test, not a better one.
Verified against the live public allowlist on 2026-08-19. Base URL is https://continuumcode.ai/v1. Keys are cont_sk_ from Settings, Account, Inference API. Personal keys need Plus or above.
curl https://continuumcode.ai/v1/chat/completions \
-H "Authorization: Bearer $CONTINUUM_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"grok-4.6","messages":[{"role":"user","content":"Review this diff."}]}'