Skip to content
Leaders Board.
Models/OpenAIClosed

GPT-5.6 Sol

#6 of all models across the language-model boards, #1 on 1 board.
#6
Overall rank
22.7
Podium score
1
Board wins
Jul 9, 2026
Released
·Chat5Reasoning12Coding1Agents and tools11Vision·Usage

Chat

Which model people prefer in blind head-to-head conversations.

15
Text Arena
1,483 · of 100
as “gpt-5.6-sol-xhigh”

Reasoning

Hard questions with checkable answers: science, math, puzzles, expert exams.

3
ARC-AGI-2
92.5% · of 78
as “GPT-5.6 Sol (Max)”
6
Epoch ECI
162.0 · of 100
as “GPT-5.6 Sol”
7
LiveBench
81.0% · of 59
as “gpt-5.6-sol-max”
11
GPQA Diamond
93.5% · of 100
as “gpt-5.6-sol_max”

Coding

Fixing real repositories, building web apps, editing code.

8
SWE-Bench Pro
82.4% · of 11
as “GPT-5.6 Sol (Codex) xhigh”
16
WebDev Arena
1,617 · of 100
as “gpt-5.6-sol-xhigh (codex-harness)”

Agents and tools

Calling functions and tools, using MCP servers, searching the web.

1
Search Arena
1,257 · of 33
as “gpt-5.6-sol-xhigh”

Vision

Understanding images, charts and documents.

7
Document Arena
1,483 · of 38
as “gpt-5.6-sol-xhigh”
12
Vision Arena
1,286 · of 100
as “gpt-5.6-sol-xhigh”

Usage

What developers actually send tokens to.

12
OpenRouter
1.9T tok · of 19
as “openai/gpt-5.6-sol-20260709”

More from OpenAI

GPT-6 SolGPT-6 AstraGPT-5.6 TerraGPT-5.5 Pre ReleaseGPT-5.5 Pro Pre ReleaseGPT-5.5 ProGPT-5.5GPT-5.4

Scores as published by each board on the capture date. Model names and logos belong to their owners; logos via logo.dev.

Weekly: the AI leaderboards with a new #1, Saturday mornings.