Xiaomi

MiMo-V2.5-Pro

Tool callingStructured outputReasoning
MiMo-V2.5-Pro launch post cover

Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution

Context
1.1M
Released
5mo ago
Parameters
—
Input / 1M tokens
$0.44
Output / 1M tokens
$0.87
Cached input / 1M tokens
$0.0036

Sources:models.devOpenRouter

Benchmarks

DeepSWE is the primary coding benchmark where an exact model evaluation exists. Other sourced evaluations appear below; missing scores are not zero.

Not evaluated in DeepSWE 1.1 as of 2026-09-22. Results are only shown for exact model versions in the official dataset. Official leaderboard.

No benchmark results have been matched to this model yet.

Specifications

Developer
Xiaomi
Model family
MiMo
Type
Language model
Predecessor
—
Successor
—
Released
April 22, 2026
Model ID
xiaomi/mimo-v2.5-pro
Openness
Open weights
License
—
Parameters
—
Input modalities
text
Output modalities
text
Modality
text->text
Tokenizer
Other
Max output tokens
131K
Knowledge cutoff
2024-12
Source freshness
Metadata synced Oct 8, 2026, 10:17 PM UTC
Additional source evidence
Source model
xiaomi/mimo-v2.5-pro
Source family
mimo
Last updated
2026-04-22
Knowledge cutoff
2024-12
Model context
1.0M
Model output limit
131K
Open weights
Yes
File attachments
No
Reasoning
Yes
Tool calling
Yes
Temperature control
Yes

Downloadable weights

Published benchmarks

BenchmarkScoreEvaluation detailsEvidence
SWE-Bench Verified78.9 · resolved—Source
SWE-Bench Pro57.2 · resolve rate2026-04-22Source
GPQA Diamond86.6 · accuracy2026-04-22Source
DeepSWE19 · scoreMiMo-V2.5 Pro comparison column · 1.1Source
ProgramBench12.5 · scoreMiMo-V2.5 Pro comparison columnSource
MiMo Code Bench40.4 · scoreMiMo-V2.5 Pro comparison columnSource
AutomationBench16 · scoreMiMo-V2.5 Pro comparison column · 1.0.6Source
Toolathlon-Verified49.1 · scoreMiMo-V2.5 Pro comparison columnSource
GDPval-AA1107 · EloMiMo-V2.5 Pro comparison column · 2.1Source
Agents' Last Exam13.2 · scoreMiMo-V2.5 Pro comparison columnSource
Terminal-Bench1.5 · scoreMiMo-V2.5 Pro comparison column · 4.0Source
Terminal-Bench65.2 · scoreMiMo-V2.5 Pro comparison column · 2.1Source
JobBench25 · scoreMiMo-V2.5 Pro comparison columnSource
CyberGym40 · scoreMiMo-V2.5 Pro comparison columnSource
MiMo Cyber Bench0 · scoreMiMo-V2.5 Pro comparison columnSource
ExploitGym0.2 · scoreMiMo-V2.5 Pro comparison columnSource
ExploitBench16.6 · scoreMiMo-V2.5 Pro comparison columnSource
SEC-Bench Pro17.7 · scoreMiMo-V2.5 Pro comparison columnSource
Complete source record (JSON)
{
  "id": "xiaomi/mimo-v2.5-pro",
  "name": "MiMo-V2.5-Pro",
  "description": "Stronger MiMo Pro tier for multimodal reasoning and coding-agent execution",
  "family": "mimo",
  "last_updated": "2026-04-22",
  "knowledge": "2024-12",
  "attachment": false,
  "reasoning": true,
  "tool_call": true,
  "temperature": true,
  "modalities": {
    "input": [
      "text"
    ],
    "output": [
      "text"
    ]
  },
  "limit": {
    "context": 1048576,
    "output": 131072
  },
  "benchmarks": [
    {
      "name": "SWE-Bench Verified",
      "score": 78.9,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5-Pro",
      "metric": "resolved"
    },
    {
      "name": "SWE-Bench Pro",
      "score": 57.2,
      "source": "https://mimo.xiaomi.com/mimo-v2-5-pro/",
      "metric": "resolve rate",
      "date": "2026-04-22"
    },
    {
      "name": "GPQA Diamond",
      "score": 86.6,
      "source": "https://mimo.xiaomi.com/mimo-v2-5-pro/",
      "metric": "accuracy",
      "date": "2026-04-22"
    },
    {
      "name": "DeepSWE",
      "score": 19,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column",
      "version": "1.1"
    },
    {
      "name": "ProgramBench",
      "score": 12.5,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column"
    },
    {
      "name": "MiMo Code Bench",
      "score": 40.4,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column"
    },
    {
      "name": "AutomationBench",
      "score": 16,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column",
      "version": "1.0.6"
    },
    {
      "name": "Toolathlon-Verified",
      "score": 49.1,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column"
    },
    {
      "name": "GDPval-AA",
      "score": 1107,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "Elo",
      "variant": "MiMo-V2.5 Pro comparison column",
      "version": "2.1"
    },
    {
      "name": "Agents' Last Exam",
      "score": 13.2,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column"
    },
    {
      "name": "Terminal-Bench",
      "score": 1.5,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column",
      "version": "4.0"
    },
    {
      "name": "Terminal-Bench",
      "score": 65.2,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column",
      "version": "2.1"
    },
    {
      "name": "JobBench",
      "score": 25,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column"
    },
    {
      "name": "CyberGym",
      "score": 40,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column"
    },
    {
      "name": "MiMo Cyber Bench",
      "score": 0,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column"
    },
    {
      "name": "ExploitGym",
      "score": 0.2,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column"
    },
    {
      "name": "ExploitBench",
      "score": 16.6,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column"
    },
    {
      "name": "SEC-Bench Pro",
      "score": 17.7,
      "source": "https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Flash-RL",
      "metric": "score",
      "variant": "MiMo-V2.5 Pro comparison column"
    }
  ],
  "release_date": "2026-04-22",
  "open_weights": true,
  "weights": [
    {
      "label": "Hugging Face",
      "url": "https://huggingface.co/XiaomiMiMo/MiMo-V2.5-Pro"
    }
  ]
}

Infrastructure

Provider availability

All providers
ProviderEndpointContextInput /M tokensOutput /M tokensCached input /M tokensCache write /M tokensUptime 1dData handling
GMICloudgmicloud/bf161.1M$0.30$0.61$0.0028—99.14%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

AtlasCloudatlas-cloud/fp81.0M$0.44$0.87$0.0036—98.23%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Xiaomixiaomi/fp81.0M$0.44$0.87$0.0036—100.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

DigitalOceandigitalocean262K$0.48$1.80$0.10—97.47%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Novitanovita1.0M$0.48$0.96$0.003956—98.03%
ZDR: Yes

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

StreamLakestreamlake1M$0.52$1.04$0.00432—60.00%
ZDR: No

Training: Unknown

Prompt logging: Unknown

Retention: Unknown

Checked 2026-10-15

Provider data:OpenRouter

Deployment graph

Can I run this?

Vertical Tabs only shows compatibility when it has structured evidence tying this exact model to hardware, runtime, quantization, memory use, and observed performance.

Browse hardware