Benchmark Almanac

MIRAGE-MMMT

GPT-4o Mar 2025 65.5% leads since June 2025
Not in model cards

Models decide whether to ask for more information or answer a grower during an agricultural consultation. They then write the question or answer.

The leaderboard reports decision accuracy against reference choices. Model judges separately score how well each question or answer serves the grower’s goal.

Released
June 2025
Kind
Multi-turn conversation
Level
University
Scored
Checked automatically
Answers
Open-ended
Tasks
All public
Measures
Capability
Language
English
In and out
Text, Images → Text
Interaction
Multi-turn
Refresh
One fixed set
Run by
Anyone
In views
Vision
Strongest model tested
Claude 3.7 Sonnet #119 on the capability index
Last model added
April 2025
Scores refreshed
November 20th, 2026
6
Models scored on its own leaderboard

GPT-4o leads with 65.5% among models released by April 2025

Best score per model from the benchmark's own leaderboard, read on November 20th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart.

The newest model with a score here came out in April 2025, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 1 GPT-4o Mar 2025, Chain-of-Thought 65.5%
  2. 2 Claude 3.7 Sonnet Chain-of-Thought 62.4%
  3. 3 Llama 4 Maverick Chain-of-Thought 59.8%
  4. 4 Llama 4 Scout Chain-of-Thought 56%
  5. 5 Llama 3.2 90B Chain-of-Thought 50.4%
  6. 6 Qwen2.5-VL-72B Chain-of-Thought 37.4%

Source: Official MIRAGE-MMMT leaderboard

#ModelScoreSettingDateSource
1 GPT-4o Mar 2025 65.5% Chain-of-Thought Mar 2025 Official leaderboard
2 Claude 3.7 Sonnet 62.4% Chain-of-Thought Feb 2025 Official leaderboard
3 Llama 4 Maverick 59.8% Chain-of-Thought Apr 2025 Official leaderboard
4 Llama 4 Scout 56% Chain-of-Thought Apr 2025 Official leaderboard
5 Llama 3.2 90B 50.4% Chain-of-Thought Sep 2024 Official leaderboard
6 Qwen2.5-VL-72B 37.4% Chain-of-Thought Jan 2025 Official leaderboard

Rankings by part

We read 2 parts from the benchmark’s own leaderboard: Clarify and Respond. Each ranks models on one part of the test. GPT-4o leads both. Arrows show how far a model moved from its overall rank among the models we track.

Clarify

6 models we track have a score here.

  1. 1 GPT-4o Mar 2025 72.8%
  2. 2 Llama 4 Maverick ↑1 69.1%
  3. 3 Llama 4 Scout ↑1 66.4%
  4. 4 Qwen2.5-VL-72B ↑2 63.9%
  5. 5 Llama 3.2 90B 56.2%
  6. 6 Claude 3.7 Sonnet ↓4 34.9%

Scores at each effort level

Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.

Lowest effort to highest, light to dark
GPT-4o
Zero-Shot
65.5%
Claude 3.7 Sonnet
Zero-Shot Chain-of-Th…
62.4%
Llama 4 Maverick
Zero-Shot Chain-of-Th…
59.8%
Llama 4 Scout
Zero-Shot Chain-of-Th…
56%
Llama 3.2 90B
Zero-Shot Chain-of-Th…
50.4%
Qwen2.5-VL-72B
Zero-Shot Chain-of-Th…
37.4%

Source: Official MIRAGE-MMMT leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
GPT-4o Mar 2025 Chain-of-Thought 65.5%
GPT-4o Mar 2025 Zero-Shot 63%
Claude 3.7 Sonnet Chain-of-Thought 62.4%
Llama 4 Maverick Chain-of-Thought 59.8%
Claude 3.7 Sonnet Zero-Shot 57.8%
Llama 4 Scout Chain-of-Thought 56%
Llama 4 Maverick Zero-Shot 53.8%
Llama 3.2 90B Chain-of-Thought 50.4%
Llama 4 Scout Zero-Shot 49.8%
Llama 3.2 90B Zero-Shot 45.1%
Qwen2.5-VL-72B Chain-of-Thought 37.4%
Qwen2.5-VL-72B Zero-Shot 31.3%

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

OpenAI
Own leaderboard Not tested since Jun 2025
GPT-4o 65.5%

Source: Official MIRAGE-MMMT leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) GPT-4o Llama 3.2 90B 0% 20% 40% 60% 80% 100% 2025 Near the ceiling (85%) GPT-4o Llama 3.2 90B

Source: Official MIRAGE-MMMT leaderboard

No lab model card reports it yet

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.

More in Instructions and chat

All 144 in Instructions and chat
  • IFEval In use Models answer about 500 prompts containing verifiable instructions such as word limits or required keywords. Code checks each instruction for strict and loose accuracy. Gemma 4 31B IT 98.9%
  • IFBench In use 58 new out-of-domain verifiable output constraints. Scored by programmatic constraint checks (strict or loose). Flint 89.2%
  • MultiChallenge In use Multi-turn conversations testing instruction retention, inference memory, versioned editing and self-coherence. Graded by LLM judge with rubrics. GPT-6.1 Sol 82%
  • Multi-IF In use Models follow verifiable instructions over three-turn conversations in eight languages, extending IFEval. Scored by instruction accuracy at each turn. LFM3-1.7B 89.1%
  • AdvancedIF In use Models respond to over 1,600 prompts with complex, multi-turn, and system-level instructions. A judge model checks each response against expert-written rubrics. MAI-Thinking-1 85%
  • MM-IFEval In use Multimodal models follow image-grounded instructions with exact output constraints. Checks mix rule-based verification and a judge model. GPT-4o 64.6%

Data audit

9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 19 months ago, in April 2025, for Llama 4 Maverick. Every benchmark

Catalog record 2 pass
  • Pass Task access All tasks are public.
  • Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (6 entries).
  • Pass Models scored 6 models we track.
  • Warning Newest score The newest came 19 months ago, in April 2025, for Llama 4 Maverick.
  • Note Frontier models None of the top 10 has a score. The strongest tested is Claude 3.7 Sonnet (#119).
  • Pass Headroom The best is 65.5%, under 85%, so it still separates models.
Leaderboard import 3 pass
  • Pass Importer We read the board on mirage-benchmark.github.io: 12 rows for models we track at the last read.
  • Pass Last read November 20th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check