Model comparison
ChatGPT vs Claude vs Gemini (2026)
GPT-6 Astra
OpenAI
$20/1M blended
1.05M context
Largest contextClaude Fable 5.1
Anthropic
$20/1M blended
1M context
Best for codingGemini 3.1 Pro (preview)
$4.50/1M blended
1M context
Cheapest to runAll three flagships now sit near a 1M-token context window, so price and specialty decide it. Gemini 3.1 Pro is the cheapest to run at $4.50/1M blended and the only one that takes audio and video. Claude Fable 5.1 ($10/$50 in/out) is the pick for coding and long agentic tasks. GPT-6 Astra has the largest context at 1.05M, but the priciest output of the three at $50/1M.
Which one to pick, by task
| Task | Best pick | Why |
|---|---|---|
| Coding / agentic work | Claude Fable 5.1 | Built for long-horizon agentic coding |
| Largest context window | GPT-6 Astra | 1.05M, the biggest of the three |
| Cheapest to run | Gemini 3.1 Pro (preview) | $4.50/1M blended (3:1 in:out) |
| Audio & video input | Gemini 3.1 Pro (preview) | The only one that natively accepts audio and video |
The coding pick reflects each model's stated design focus, not a benchmark score. We don't publish scores we haven't measured.
The three flagships, head to head
List pay-as-you-go API prices, USD per 1M tokens (prompts up to ~200K).
| Model | Context | Input /1M | Output /1M | Blended /1M | Modality | Weights |
|---|---|---|---|---|---|---|
| GPT-6 AstraOpenAI | 1.05M | $10 | $50 | $20 | Text + image in | Proprietary |
| Claude Fable 5.1Anthropic | 1M | $10 | $50 | $20 | Text + image in | Proprietary |
| Gemini 3.1 Pro (preview)Google | 1M | $2 | $12 | $4.50 | Multimodal (text/image/audio/video) | Proprietary |
Blended = a 3:1 input:output mix (assistant/agent workloads re-send a lot of context), shown alongside the raw rates, never on its own.
Cheaper tiers from each maker
Every family sells a lighter, cheaper model for high-volume or simpler work.
| Model | Context | Input /1M | Output /1M | Blended /1M | Modality | Weights |
|---|---|---|---|---|---|---|
| OpenAI | ||||||
| GPT-6 AstraOpenAI | 1.05M | $10 | $50 | $20 | Text + image in | Proprietary |
| GPT-5.6 TerraOpenAI | 1.05M | $2 | $12 | $4.50 | Text + image in | Proprietary |
| Anthropic | ||||||
| Claude Fable 5.1Anthropic | 1M | $10 | $50 | $20 | Text + image in | Proprietary |
| Claude Sonnet 5.5Anthropic | 1M | $2 | $10 | $4 | Text + image in | Proprietary |
| Gemini 3.1 Pro (preview)Google | 1M | $2 | $12 | $4.50 | Multimodal (text/image/audio/video) | Proprietary |
| Gemini 3.8 FlashGoogle | 1M | $0.75 | $3.75 | $1.50 | Multimodal (text/image/audio/video) | Proprietary |
Just want the app? The ~$20/mo consumer plans
Most people comparing these three want the chatbot subscription, not the API. Here are the mainstream paid tiers side by side.
| Plan | Price / mo | What stands out | Best for |
|---|---|---|---|
| ChatGPT PlusOpenAI ChatGPT | $20 | Full frontier feature set at the mainstream price | Most paying individuals |
| Claude ProAnthropic Claude | $20 | Bundles Claude Code (terminal coding agent) | Devs, writers, analysts |
| Google AI ProGoogle Gemini | $19.99 | Bundles 5 TB storage + YouTube perks | Mainstream value pick |
Consumer-plan prices verified June 2026; usage limits change often. Compare every plan (free → $200) →
The honest verdict
There is no single winner, and any comparison that claims one is selling something. The practical split is Claude Fable 5.1 for coding and agentic work, Gemini 3.1 Pro for the lowest cost and for audio, image, and video input, and GPT-6 Astra as the largest-context generalist. If the deciding factor is the monthly bill, Gemini 3.1 Pro wins on the numbers at $4.50/1M blended.
FAQ
Which is cheapest: ChatGPT, Claude, or Gemini?
Gemini 3.1 Pro is the cheapest of the three to run, at $4.50 per 1M tokens blended ($2 input / $12 output), well under Claude Fable 5.1 and GPT-6 Astra.
Which is best for coding?
Claude Fable 5.1. It is built for long-horizon agentic coding, which makes it the pick of the three for writing and fixing code.
Which has the biggest context window?
GPT-6 Astra, at 1.05M. That is slightly ahead of the other two, which are 1M.
Which one handles images, audio, and video?
Gemini 3.1 Pro is the only one of the three whose inputs include audio and video, not just text and images.
Which $20-a-month plan is best: ChatGPT Plus, Claude Pro, or Google AI Pro?
All three mainstream chatbot subscriptions cost about $20/month (ChatGPT Plus at $20, Claude Pro at $20, Google AI Pro at $19.99). Claude Pro is the pick if you code (it bundles Claude Code) while ChatGPT Plus has the broadest feature set and Google AI Pro is the value option, bundling 5 TB of storage and YouTube perks. Usage limits change often; prices verified June 2026.
Put it to work
Primary sources: OpenAI's official pricing ↗Anthropic's official pricing ↗Google's official pricing ↗
API prices from BitByteCore's hand-verified Frontier Ledger (v2026-10-01, last updated 1 Oct 2026); consumer-plan prices verified June 2026. Pricing moves fast, so confirm with each provider before you rely on a figure. This comparison updates when the Ledger does.