Skip to main content

Frontier Model Comparison

Link copied!

A dated snapshot of the current top-tier, closed ("frontier") models, meaning the API-only models from the major labs that set the capability bar. Open-weight models (DeepSeek, Kimi, Qwen, Mistral, GLM) are in Open-Weight Models. For their hosted API prices, check OpenRouter. It's here for the at-a-glance shape of the field, not as a live price sheet: prices and versions move weekly, so use the leaderboards for current specs, speed, and prices. As of July 19, 2026. Prices are standard-tier API, USD per 1M tokens, short-context base rate (long-context tiers typically cost ~2×). Output speed (tok/s) is essentially never an official spec, so check Artificial Analysis for measured throughput. Update (July 21, 2026): Google shipped Gemini 3.6 Flash (reflected below) alongside Gemini 3.5 Flash-Lite (cheaper, faster) and a security-only 3.5 Flash Cyber; Gemini 3.5 Pro is still in partner testing and behind schedule, and Gemini 4 pre-training has begun.

Model Provider Released Context Modalities (in) Reasoning In $/1M Out $/1M
GPT-5.6 "Sol" (flagship) OpenAI Jul 2026 1M text, image Yes $5.00 $30.00
GPT-5.6 "Terra" (balanced) OpenAI Jul 2026 ~1M text, image Yes $2.50 $15.00
GPT-5.6 "Luna" (fast) OpenAI Jul 2026 ~1M text, image Yes $1.00 $6.00
Claude Fable 5 (flagship) Anthropic 2026 1M text, image Yes $10.00 $50.00
Claude Opus 4.8 Anthropic 2026 1M text, image Yes $5.00 $25.00
Claude Sonnet 5 Anthropic 2026 1M text, image Yes $2.00 ¹ $10.00 ¹
Claude Haiku 4.5 Anthropic 2026 1M text, image Yes $1.00 $5.00
Gemini 3.1 Pro (flagship) Google Feb 2026 ~1M text, image, audio, video Yes ~$2.00 ² ~$12.00 ²
Gemini 3.6 Flash Google Jul 2026 ~1M text, image, audio, video Yes ~$1.50 ~$7.50
Grok 4.5 xAI 2026 500K text, image Yes $2.00 $6.00
Command A Plus Cohere May 2026 128K text No $2.50 $10.00
Amazon Nova 2 Pro Amazon 2026 1M multimodal n/a ~$2.50 ~$12.50

¹ Sonnet 5 price rises to $3 / $15 on Sept 1, 2026.
² Gemini 3.1 Pro is tiered by context length (≤200K shown). Verify on the Gemini pricing page.

Most flagships are now hybrid reasoning models (a toggleable "thinking" mode). Sources: OpenAI · Anthropic · Google · xAI.

Live sources (bookmark these instead of trusting the table for exact numbers): Artificial Analysis (intelligence + speed + price + latency), OpenRouter models (live per-provider pricing & real throughput), llm-stats.com (spec sheets with pricing baked in).

Link copied!