Task results
Spotting furniture assembly mistakes (Furniture Assembly)
Given a photo of an IKEA furniture build and the official assembly manual, the model decides whether the build is correct and, if not, which step was done wrong.
How to read this table
- Higher rows are rated higher. A range like “1–2” means the gap is within the margin of error.
- The score is the share of a shared set of problems the model solved (%).
- ± is how far the score moves when re-measured, so small gaps do not mean much.
- Prices are a guide per million tokens, reading then writing.
- Length is how much text the model can read at once (tokens).
± range of the top model
| Rank | Model | Developer | Accuracy (%) | Evaluated | Input $ /1M | Output $ /1M | Context | Supports | Weights | Released | Price and specs |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1–6 | Claude Opus 5.5 | Anthropic | Rating 76.2 ±13.2 | $4 | $20 | 1M | Anthropic Released 2026-09 Price $4 / $20 Length 1M | ||||
| 1–6 | GPT-6.1 Sol | OpenAI | Rating 71.4 ±13.8 | $2 | $10 | 1.05M | OpenAI Released 2026-09 Price $2 / $10 Length 1.05M | ||||
| 1–6 | GPT-6 Astra | OpenAI | Rating 71.4 ±13.8 | $10 | $50 | 1.05M | OpenAI Released 2026-09 Price $10 / $50 Length 1.05M | ||||
| 1–7 | Claude Sonnet 5.5 | Anthropic | Rating 64.3 ±15.4 | $2 | $10 | 1M | Anthropic Released 2026-09 Price $2 / $10 Length 1M | ||||
| 1–9 | Gemini 4 Argon | Rating 60.7 ±15.8 | $2 | $10 | 1.05M | Google Released 2026-09 Price $2 / $10 Length 1.05M | |||||
| 1–11 | Claude Fable 5.1 | Anthropic | Rating 57.1 ±16.0 | $10 | $50 | 1M | Anthropic Released 2026-09 Price $10 / $50 Length 1M | ||||
| 4–14 | Claude Opus 5 | Anthropic | Rating 44.0 ±17.3 | $5 | $25 | 1M | Anthropic Released 2026-07 Price $5 / $25 Length 1M | ||||
| 5–16 | GPT-6 Sol | OpenAI | Rating 40.5 ±18.0 | $2 | $10 | 1.05M | OpenAI Released 2026-09 Price $2 / $10 Length 1.05M | ||||
| 5–17 | GPT-5.6 Sol | OpenAI | Rating 38.1 ±17.8 | $4 | $20 | 1.05M | OpenAI Released 2026-07 Price $4 / $20 Length 1.05M | ||||
| 6–19 | GPT-5.6 Terra | OpenAI | Rating 34.5 ±18.0 | $2 | $12 | 1.05M | OpenAI Released 2026-07 Price $2 / $12 Length 1.05M | ||||
| 6–19 | Muse Wonder | Meta | Rating 34.5 ±18.0 | $4 | $20 | 1.05M | Meta Released 2026-10 Price $4 / $20 Length 1.05M | ||||
| 7–24 | Claude Haiku 5.5 | Anthropic | Rating 25.0 ±18.1 | $0.1 | $0.5 | 1M | Anthropic Released 2026-10 Price $0.1 / $0.5 Length 1M | ||||
| 7–26 | GPT-5.5 | OpenAI | Rating 20.2 ±17.9 | $5 | $30 | 1.05M | OpenAI Released 2026-04 Price $5 / $30 Length 1.05M | ||||
| 7–26 | GPT-6 Luna | OpenAI | Rating 20.2 ±17.9 | $0.1 | $0.5 | 1.05M | OpenAI Released 2026-09 Price $0.1 / $0.5 Length 1.05M | ||||
| 8–28 | Claude Opus 4.8 | Anthropic | Rating 17.9 ±17.9 | $5 | $25 | 1M | Anthropic Released 2026-05 Price $5 / $25 Length 1M | ||||
| 8–28 | GPT-5.6 Luna | OpenAI | Rating 17.9 ±17.9 | $0.2 | $1.2 | 1.05M | OpenAI Released 2026-07 Price $0.2 / $1.2 Length 1.05M | ||||
| 9–28 | Grok 4.6 | xAI | Rating 14.3 ±17.5 | $2 | $6 | 500K | xAI Released 2026-08 Price $2 / $6 Length 500K | ||||
| 10–30 | GPT-5.2 | OpenAI | Rating 11.9 ±17.7 | $1.75 | $14 | 400K | OpenAI Released 2025-12 Price $1.75 / $14 Length 400K | ||||
| 10–31 | GPT-5.4 | OpenAI | Rating 10.7 ±17.5 | $2.5 | $15 | 1.05M | OpenAI Released 2026-03 Price $2.5 / $15 Length 1.05M | ||||
| 12–34 | Claude Fable 5 | Anthropic | Rating 8.3 ±17.3 | $10 | $50 | 1M | Anthropic Released 2026-06 Price $10 / $50 Length 1M | ||||
| 12–34 | DeepSeek-V4.1 Flash | DeepSeek | Rating 6.0 ±16.8 | — | — | 1M | DeepSeek Released 2026-09 Price — Length 1M | ||||
| 12–34 | Kimi K3 | Moonshot | Rating 6.0 ±17.1 | $3 | $15 | 1.05M | Moonshot Released 2026-07 Price $3 / $15 Length 1.05M | ||||
| 12–34 | Claude Opus 4.7 | Anthropic | Rating 4.8 ±17.2 | $5 | $25 | 1M | Anthropic Released 2026-04 Price $5 / $25 Length 1M | ||||
| 12–34 | Gemini 3.8 Flash | Rating 2.4 ±16.6 | $0.75 | $3.75 | 1.05M | Google Released 2026-09 Price $0.75 / $3.75 Length 1.05M | |||||
| 13–34 | Claude Opus 4.5 | Anthropic | Rating 0.0 ±16.4 | $5 | $25 | 200K | Anthropic Released 2025-11 Price $5 / $25 Length 200K | ||||
| 13–34 | Claude Opus 4.6 | Anthropic | Rating 0.0 ±16.4 | $5 | $25 | 1M | Anthropic Released 2026-02 Price $5 / $25 Length 1M | ||||
| 15–34 | Gemini 3.1 Pro (preview) | Rating 0.0 ±15.8 | $2 | $12 | 1.05M | Google Released 2026-02 Price $2 / $12 Length 1.05M | |||||
| 15–34 | Gemini 3.7 Flash | Rating 0.0 ±15.8 | $0.75 | $3.75 | 1.05M | Google Released 2026-08 Price $0.75 / $3.75 Length 1.05M | |||||
| 18–34 | Gemini 3.6 Flash | Rating 0.0 ±15.1 | $0.75 | $3.75 | 1.05M | Google Released 2026-07 Price $0.75 / $3.75 Length 1.05M | |||||
| 18–34 | Grok 4.5 | xAI | Rating 0.0 ±15.0 | $2 | $6 | 500K | xAI Released 2026-07 Price $2 / $6 Length 500K | ||||
| 19–34 | Kimi K2.6 | Moonshot | Rating 0.0 ±14.6 | $0.95 | $4 | 262.1K | Moonshot Released 2026-04 Price $0.95 / $4 Length 262.1K | ||||
| 20–34 | Grok 4.7 | xAI | Rating 0.0 ±14.6 | $2 | $6 | 500K | xAI Released 2026-09 Price $2 / $6 Length 500K | ||||
| 20–34 | Mistral Large 4 | Mistral | Rating 0.0 ±14.6 | $1.36 | $4.18 | 524.3K | Mistral Released 2026-10 Price $1.36 / $4.18 Length 524.3K | ||||
| 20–34 | Qwen3.8 Max (2026-09) | Alibaba | Rating 0.0 ±14.6 | — | — | 1M | Alibaba Released 2026-09 Price — Length 1M |
Best accuracy over time
2025-11-24Release date2026-10-12
How to read this page
- It tests spatial reasoning that connects the manual's diagrams to the real object. The value is accuracy.
- The accuracy is the share of problems solved, with the part you could get by guessing removed.
- In the Supports column the icons are, from left: can use tools / can read images / can answer in a fixed format (JSON); faded means not supported or unknown.
- Open-weights models have their contents (the weights) published, so you can also run them on your own computer or server. Models without the open-weights mark are normally used through the developer's own services (apps or API); for some models the weights status is unknown.
- Names such as ChatGPT next to a model are the consumer services its developer offers. Check on the service itself whether that model is available there.
Data: Arena, Epoch AI, Swallow LLM team (CC BY 4.0), Models.dev (MIT), updated 2026-10-24, data sources