DeepSWE
No official DeepSWE mini-swe-agent row for this identity on the public board we fetched.
Not listed as a Continuum hosted id. MIT on the official repo.
HF config.json (fetched 19 Aug 2026): MiMoV2ForCausalLM, 48 layers, hidden 4096, 256 routed experts, 8 experts/token, YaRN window 1,048,576. Vision config is present. FP8 quantization_config on the Hub card. Chat template is present.
Use it when you want Xiaomi’s open multimodal checkpoint. AA 38 and vals 39.91% are cited; we do not invent a DeepSWE percent.
No official DeepSWE mini-swe-agent row for this identity on the public board we fetched.
vals.ai MiMo-V2.5 card, fetched 19 Aug 2026: Index 39.91% ±1.31, $0.061 per Index test, 18 min 13 s latency. Those are Index-page extras, not DeepSWE, and not hero chips. No Updates prose with named non-zero benches was present on that HTML. The Accuracy Rankings bars printed 0.0% placeholders: 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 |
|---|---|---|---|---|---|
| AA list pair | $0.14 / $0.28 per 1M | Printed AA pricing sentence. Matches the OpenRouter catalog pair. | Artificial Analysis MiMo-V2.5 | source | 2026-08-19 |
| AA throughput | 58.4 tok/s | Printed Speed row. Not a coding score. | Artificial Analysis MiMo-V2.5 | source | 2026-08-19 |
| AA Index eval cost | $25.18 | Printed “it cost $25.18 to evaluate MiMo-V2.5 on the Intelligence Index.” | Artificial Analysis MiMo-V2.5 | source | 2026-08-19 |
| vals Index extras | $0.061/test · 18 min 13 s | Printed Cost / Test and Latency on the vals hero. Not DeepSWE. | vals.ai MiMo-V2.5 | source | 2026-08-19 |
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.
Community quants: Community quants are community.
This is the model repo, not Download for Mac. Downloads and likes from the Hugging Face API on 2026-08-19.
usedStorage and the sum of .safetensors sizes can differ (LFS pointers, non-shard files, duplicate copies). Cited separately. Source: recursive tree API, 2026-08-19.
| Path | Bytes |
|---|---|
audio_tokenizer/model.safetensors | 652,622,472 |
model_mtp.safetensors | 1,189,405,960 |
model_pp0_ep0_shard0.safetensors | 34,369,159,872 |
model_pp0_ep0_shard1.safetensors | 14,463,302,008 |
model_pp0_ep1_shard0.safetensors | 34,369,162,432 |
model_pp0_ep1_shard1.safetensors | 3,490,619,024 |
model_pp0_ep2_shard0.safetensors | 34,369,162,432 |
model_pp0_ep2_shard1.safetensors | 3,490,619,024 |
model_pp0_ep3_shard0.safetensors | 34,369,169,600 |
model_pp0_ep3_shard1.safetensors | 3,490,619,752 |
model_pp0_ep4_shard0.safetensors | 34,369,170,624 |
model_pp0_ep4_shard1.safetensors | 3,490,619,856 |
Showing the first 12 of 18 safetensor shards. The tree also holds 21 non-shard files, none of them listed here. The snapshot was capped at 12 shards when the tree API was read on 2026-08-19, so the remaining 6 are on the Files tab, not on this page. Totals above are the full tree API sum, not this slice. Byte sizes are Hub tree API size fields, never invented.
Not listed as a Continuum hosted id. OpenRouter xiaomi/mimo-v2.5.
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.
OpenRouter catalog 19 Aug 2026: $0.14 / $0.28 per million, cache read $0.0028 /M, created 1776874269, context 1,050,000 / 131,072. No DeepSWE $/task.
Artificial Analysis MiMo-V2.5 page (19 Aug 2026) prints the same $0.14 / $0.28 pair, 58.4 tok/s, and “it cost $25.18 to evaluate MiMo-V2.5 on the Intelligence Index.”
Artificial Analysis Intelligence Index 38, fetched 19 Aug 2026. Same page prints $0.14 / $0.28, 58.4 tok/s, and $25.18 to run the Index. The AA tile is the published integer 38. AA did not print a DeepSWE-style CI or step count we will copy onto the chip.
| Claim | Source | As of |
|---|---|---|
| OR catalog $0.14 / $0.28, cache read $0.0028/M, 1,050,000 / 131,072, created 1776874269 | OpenRouter catalog | 2026-08-19 |
| 5.46T week tokens | OpenRouter rankings This Week | 2026-08-19 |
| HF safetensors.total 310,775,040,000; usedStorage 315,693,598,475; 18 safetensor files; likes 400; created 2026-04-27T13:37:38Z; MIT | Hugging Face API XiaomiMiMo/MiMo-V2.5 | 2026-08-19 |
| AA Index 38; printed $0.14 / $0.28; 58.4 tok/s; $25.18 to evaluate Index | Artificial Analysis MiMo-V2.5 | 2026-08-19 |
| vals Index 39.91% ±1.31; $0.061/test; 18 min 13 s | vals.ai MiMo-V2.5 | 2026-08-19 |
| OpenRouter id xiaomi/mimo-v2.5, context 1,050,000 | OpenRouter /api/v1/models | 2026-08-19 |
| HF downloads 426,872 | Hugging Face API XiaomiMiMo/MiMo-V2.5 | 2026-08-19 |
| 18 safetensor shards; shard bytes 315,693,004,496; tree files 39 | Hugging Face tree API XiaomiMiMo/MiMo-V2.5 | 2026-08-19 |
| Spaces API returned 2; official collection MiMo-V2.5 (4 items, 91 upvotes) | Hugging Face Spaces / collections API | 2026-08-19 |
No other card in this cluster shares a published DeepSWE official row we can put next to this one. We will not compare on vendor-blog numbers.
Continue through Xiaomi's model family with MiMo-V2.5-Pro.
The official DeepSWE board updated 13 Aug 2026 did not list MiMo-V2.5. Omitting is the rule.
Different harnesses. We cite both integers and do not average them.
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.
No Continuum hosted id is verified for this model, so this page does not invent a curl target. Use the first-party API or the OpenRouter slug xiaomi/mimo-v2.5.