Hard, high-stakes work
Start with each company's flagship model and test it on your real task.
GPT-5.6 Sol · Claude Opus 5 · Grok 4.5
Explore AI
Find the right model faster.
Pick the kind of work you have, compare a few sensible options, then use your own task results to decide what earns a place in your stack.
Daily balanced work
Use the middle tier when you want strong results without flagship cost.
GPT-5.6 Terra · Claude Sonnet 5 · Grok 4.3
Fast, repeatable work
Choose a low-cost model for volume, drafts, routing, and sub-agents.
GPT-5.6 Luna · Gemini Flash-Lite · Grok Build
Two labels worth knowing
Working memory
What “1M context” means
It can keep up to one million tokens in a single request. That total includes your prompt, files, chat history, and the model’s reply.
Very roughly, 1M tokens is around 750,000 words. More memory helps with large codebases and long documents.
API price
What “input / output” means
Input is everything you send to the model. Output is everything it generates back. Each price is the cost per one million tokens.
$1.50 in · $7.50 out means generated text costs more than the text you send. These are API prices, not app subscription prices.
A
Compare all companies Anthropic
6 models to compare
A
FeaturedAnthropic
Claude Opus 4.8
Anthropic's most capable model for demanding agentic work.
Working memory
1M tokens
How much it can handle at once
API price / 1M
$5 in · $25 out
You send it · it writes back
Long tasksCodingCareful reasoning
Verified 2026-07-27Official source
A
FeaturedAnthropic
Claude Opus 5
Built for complex agentic coding and long-horizon enterprise work.
Working memory
1M tokens
How much it can handle at once
API price / 1M
$5 in · $25 out
You send it · it writes back
Agentic codingLong tasksDeep reasoning
Verified 2026-07-28Official source
A
Anthropic
Claude Fable 5
Anthropic's most capable model, for demanding reasoning and long-horizon agentic work.
Working memory
1M tokens
How much it can handle at once
API price / 1M
$10 in · $50 out
You send it · it writes back
Deep reasoningLong-horizon agentsComplex coding
Verified 2026-07-31Official source
A
Anthropic
Claude Sonnet 5
A fast frontier model balancing intelligence, speed, and cost.
Working memory
1M tokens
How much it can handle at once
API price / 1M
$2 in · $10 out
You send it · it writes back
CodingData analysisFast agents
Verified 2026-07-28Official source
A
Anthropic
Claude Haiku 4.5
Anthropic's fastest option for high-volume, cost-sensitive work.
Working memory
200K tokens
How much it can handle at once
API price / 1M
$1 in · $5 out
You send it · it writes back
SpeedHigh volumeSub-agents
Verified 2026-07-28Official source
A
Anthropic
Claude Sonnet 4.6
A strong balance of intelligence, speed, and practical cost.
Working memory
1M tokens
How much it can handle at once
API price / 1M
$3 in · $15 out
You send it · it writes back
DebuggingImplementationWriting
Verified 2026-07-27Official source
GGoogle
Compare all companies 3 models to compare
GGoogle
FeaturedGemini 3.6 Flash
Google's intelligent, speed-focused model with search and grounding.
Working memory
1M tokens
How much it can handle at once
API price / 1M
$1.5 in · $7.5 out
You send it · it writes back
SpeedMultimodalGrounding
Verified 2026-07-27Official source
GGoogle
Gemini 3.1 Flash-Lite
Google's cost-efficient model for high-volume agentic tasks.
Working memory
1M tokens
How much it can handle at once
API price / 1M
$0.25 in · $1.5 out
You send it · it writes back
Low costHigh volumeTranslation
Verified 2026-07-28Official source
GGoogle
Gemini 2.5 Pro
A multipurpose reasoning model for coding and complex work.
Working memory
1.05M tokens
How much it can handle at once
API price / 1M
$1.25 in · $10 out
You send it · it writes back
Complex reasoningCodingLong context
Verified 2026-07-28Official source
O
Compare all companies OpenAI
3 models to compare
O
FeaturedOpenAI
GPT-5.6 Sol
Frontier model for complex professional reasoning and coding.
Working memory
1.05M tokens
How much it can handle at once
API price / 1M
$5 in · $30 out
You send it · it writes back
Complex codingAgentsDeep reasoning
Verified 2026-07-27Official source
O
OpenAI
GPT-5.6 Terra
Balanced frontier intelligence for cost-aware production work.
Working memory
1.05M tokens
How much it can handle at once
API price / 1M
$2.5 in · $15 out
You send it · it writes back
Balanced valueCodingReasoning
Verified 2026-07-27Official source
O
OpenAI
GPT-5.6 Luna
Cost-sensitive model for high-volume production workloads.
Working memory
1.05M tokens
How much it can handle at once
API price / 1M
$1 in · $6 out
You send it · it writes back
High volumeFast iterationLow cost
Verified 2026-07-27Official source
x
Compare all companies xAI
3 models to compare
x
FeaturedxAI
Grok 4.5
xAI's flagship model for reasoning with a large context window.
Working memory
500K tokens
How much it can handle at once
API price / 1M
$2 in · $6 out
You send it · it writes back
ReasoningCurrent informationLarge context
Verified 2026-07-27Official source
x
xAI
Grok 4.3
A lower-cost reasoning model with a one-million-token context.
Working memory
1M tokens
How much it can handle at once
API price / 1M
$1.25 in · $2.5 out
You send it · it writes back
ValueLarge contextReasoning
Verified 2026-07-28Official source
x
xAI
Grok Build 0.1
A coding-focused model for building and editing software.
Working memory
256K tokens
How much it can handle at once
API price / 1M
$1 in · $2 out
You send it · it writes back
CodingImplementationLow cost
Verified 2026-07-28Official source