Model · stat sheet
Xiaomi MiMo 2.6 Flash
The cheap, fast end of the V2.6 series: 1M context at $0.14/$0.28 per 1M tokens. Early read — the value column is unrated until an index lands.
What it is
The efficiency end of Xiaomi's September 2026 MiMo-V2.6 series: same 1M-token context window as the Pro, a quarter of the price ($0.14 / $0.28 per 1M tokens), and speed as the headline feature. Xiaomi's release chart (Sept 2026, vendor-reported) puts it at 28.8 on Terminal Bench 4.0 — below the Pro's 34.9, which matches how it scores here. There is no published Artificial Analysis index for it yet, so the value column reads unrated.
How the composite is built
Blended from six axes (Coding 24% · Terminal 20% · Reasoning 20% · Tool use 15% · Context 13% · Speed 8%), it lands at 82 — level with last generation's Pro.
Coding — 82
measured on real edit-and-run workloads: multi-file changes, tests passing, diffs that hold up under review.
Terminal — 78
measured on long agentic sessions in a shell: planning, running tools, recovering from failed commands.
Reasoning — 83
measured on planning and hard decompositions — the questions a model has to answer before it acts.
Tool use — 80
measured on calling tools correctly: right arguments, right ordering, no hallucinated functions.
Context — 85
measured on usable context under load: what survives at the top of a long session.
Speed — 92
throughput and latency; higher is faster.
Where it fits
High-volume routing: cheap subagents, CI triage, summarising, the 80% of calls inside an agent loop that don't need the strongest model in the fleet. With 1M of context at $0.175 blended, it is one of the few models on the board where "just send the whole log" is a sane strategy.
Limits
Early read, scored 26 Sep 2026. Terminal work is its weakest axis (78) — this is not the model for a five-hour unattended refactor. No AA index yet, so its value rating is a blank, not a verdict.
Price and access
$0.14 in / $0.28 out per 1M tokens ($0.175 blended). Open weights; also served via OpenRouter and OpenCode Zen/Go.
Alternatives on this board
- Xiaomi MiMo 2.6 Pro — 4 composite points more, 3× the price.
- GPT-5.6 Luna — the closed-model alternative at the same composite.
- Xiaomi MiMo 2.5 Pro — previous generation, same overall score.
Sources
- OpenRouter: MiMo-V2.6-Flash — pricing as of Sept 2026.
- Xiaomi MiMo-V2.6 release materials (Sept 2026) — benchmark figures are vendor-reported.
- OpenCode Zen / OpenCode Go listings, Sept 2026.
Scores are fullauto.online's composite index (0–100): Coding 24% · Terminal 20% · Reasoning 20% · Tool use 15% · Context 13% · Speed 8%. Editorial, not a vendor benchmark; 2026 tiers are early reads. Last scored 26 Sep 2026 · back to the leaderboard.