LLM Reference

The data & sql leaderboard · for knowledge workers

Best for data & sql

4 editor picks · 7 eligible models · Reads schemas, writes queries that actually run.

Editorial pick plus benchmark and API pricing context.

See raw /best
EDITOR'S CHOICEResearched 23d ago

GPT-5.5

OpenAI · 1.05m context
Excellent

Reads schemas and writes queries that actually run.

Best text-to-SQL accuracy in production — HumanEval 94.2 and top-tier reasoning pick the right join and respect dialect quirks.

Open model
The numbers
$/1M out
$30.00
$5.00 input
Context
1.05m
max window
Pros
  • +Top SQL accuracy & reasoning
  • +Good dialect awareness
  • +1M context for big schemas
Cons
  • −$30 / 1M out — drop to DeepSeek V4 Pro for high volume

Also worth picking

The runners-up

ranked by editorial pick orderEditorial tiersExcellentStrongSolid
#2
Claude Sonnet 4.6
Anthropic · 1m
$15.00 / 1M out
Best at reasoning over messy schemas with hundreds of tables.
#3
DeepSeek V4 Pro
DeepSeek · 1m
$0.87 / 1M out
LiveCodeBench leader at $0.87 out — excellent SQL value, open weights.
#4
Gemini 3 Pro
Google DeepMind · 1m
$5.00 / 1M out
A 1M window fits the whole schema for one-shot exploration.

Eligibility

7 models are eligible for this board

Eligibility means tagged with useCases: [data-sql]. Pins must come from this pool.

All picks