Direct answer
Credits = tokens × the model's price × 100
Since June 1, 2026, GitHub Copilot bills chat, agents and the CLI in GitHub AI Credits. One credit is $0.01. Each request costs its input, cached, cache-write and output tokens at the model's per-1M-token rate. Pro includes 1,500 credits a month, Pro+ 7,000 and Max 20,000. Business adds 1,900 and Enterprise 3,900 per user to a shared pool. The allowance resets at 00:00 UTC on the 1st, and extra usage costs $0.01 per credit up to your budget. Code completions and next edit suggestions are free. Checked November 8, 2026.
Estimate your Copilot AI Credits for a month
Pick a plan, then enter tokens per day for each model you use. Rates are GitHub's published per-1M-token prices, checked November 8, 2026.
AI credits used this month
8,800 credits
$88.00 of model usage
Share of your allowance
126%
7,000 included credits on Copilot Pro+
Allowance runs out
Day 24
of a 30-day month at an even 293.3 credits per calendar day.
Overage at this pace
$18.00
1,800 credits beyond the included allowance, at $0.01 per credit.
Additional usage budget
$10.00 = 1,000 credits
Runs out on day 28. 800 credits more would be blocked if "Stop usage when budget limit is reached" is on.
By model, for you
- Claude Sonnet 5.58,800 credits
400 credits per day × 22 days = 8,800 credits
Load a usage preset
Illustrative daily token volumes, not GitHub data. They replace the model rows and keep your plan, budget and days.
Check GitHub's own example
GitHub's budgets doc: 400 Business licenses cost $7,600 a month and fund a 760,000-credit pool. An enforced $5,000 enterprise budget allows up to $12,600 in combined license and metered charges, not $5,000.
Transparent math
How Copilot AI Credits are calculated
GitHub prices every model per 1M tokens in dollars and converts the total to credits at $0.01 each. The calculator does the same for each model row, then compares the month's total with your plan's allowance.
June 1, 2026
How AI Credits replaced premium requests
Under the old system each chat or agent prompt cost a fixed number of premium requests: one request times the model's multiplier, whatever the size of the context. Pro came with 300 a month, Pro+ with 1,500, and extra requests cost $0.04 each. On June 1, 2026 GitHub moved every plan to usage-based billing: the model and the number of tokens now set the price. Only Pro and Pro+ subscribers on an existing annual plan stay on premium requests until that plan ends. The full background is in GitHub Copilot AI Credits explained and GitHub Copilot billing in 2026.
| Premium requests (legacy) | AI Credits (since June 1, 2026) | |
|---|---|---|
| Unit | 1 prompt × model multiplier | tokens × model price, $0.01 per credit |
| Pro monthly allowance | 300 premium requests | 1,500 credits |
| Pro+ monthly allowance | 1,500 premium requests | 7,000 credits |
| Extra usage | $0.04 per premium request | $0.01 per credit, within your budget |
| Long agent run | same price per prompt | grows with every token re-sent |
Dollar equivalence only: $0.04, the old price of one extra premium request, buys 4 AI credits. GitHub publishes no conversion rate between the two units.
Example dated November 8, 2026
Worked scenario: a Pro+ developer in agent mode
One developer uses Claude Sonnet 5.5 in agent mode on 22 workdays of a 30-day month, with a $10 additional-usage budget. The token volumes are illustrative; the rates are GitHub's.
| Bucket | Tokens/day | $ per 1M | Credits/day |
|---|---|---|---|
| Input | 600,000 | $2.00 | 120 |
| Cached input | 6,000,000 | $0.10 | 60 |
| Cache write | 400,000 | $2.50 | 100 |
| Output | 120,000 | $10.00 | 120 |
| Per day | 7,120,000 | $4.00 | 400 |
Month (22 days)
8,800 credits
Of Pro+ 7,000
126%
Runs out
Day 24
Overage
1,800 = $18
At an even 293.3 credits per calendar day, the 7,000 included credits are gone on day 24. The $10 budget covers 1,000 more and runs out on day 28, leaving 800 credits that would be blocked if the budget stops usage. Two cheaper fixes: with auto model selection the same work costs 7,920 credits, so the $9.20 overage fits the budget; or Max (20,000 credits) puts the month at 44% of the allowance, for $61 more than Pro+.
Organizations
400 Business licenses and a $5,000 budget
This is GitHub's own example from its budgets doc. 400 licenses × $19 = $7,600 a month, and 400 × 1,900 = 760,000 pooled credits. An enforced $5,000 enterprise budget adds up to 500,000 metered credits, so the month can absorb 1,260,000 credits and cost up to $12,600: $7,600 in licenses plus $5,000 of usage. The budget never caps the license fees.
Pooled vs per user
Why idle seats still help
Every license funds the pool, used or not. If only 300 of those 400 people use chat or agents, each active user can draw about 2,533 credits before overage, not 1,900. At 3,000 credits each, the org uses 900,000 credits, the pool empties on day 26 of a 30-day month, and the 140,000-credit overage costs $1,400, inside the $5,000 budget.
Rates used
Copilot model rates per 1M tokens
From GitHub's “Models and pricing for GitHub Copilot”, checked November 8, 2026. Multiply a dollar rate by 100 for credits per 1M tokens. A dash means GitHub lists no cache write cost for that model.
Show all 41 text-model rates
| Model | Input | Cached input | Cache write | Output |
|---|---|---|---|---|
| GPT-5.3-Codex | $1.75 | $0.175 | — | $14.00 |
| GPT-5.4 nano | $0.20 | $0.02 | — | $1.25 |
| GPT-5.6 Luna≤ 200K input tokens | $0.20 | $0.02 | $0.25 | $1.20 |
| GPT-5.6 Luna (long context)> 200K input tokens | $0.40 | $0.04 | $0.50 | $1.80 |
| GPT-5.6 Sol≤ 272K input tokens | $4.00 | $0.40 | $5.00 | $20.00 |
| GPT-5.6 Sol (long context)> 272K input tokens | $8.00 | $0.80 | $10.00 | $30.00 |
| GPT-5.6 Terra≤ 272K input tokens | $2.00 | $0.20 | $2.50 | $12.00 |
| GPT-5.6 Terra (long context)> 272K input tokens | $4.00 | $0.40 | $5.00 | $18.00 |
| GPT-6 Astra≤ 272K input tokens | $10.00 | $1.00 | $12.50 | $50.00 |
| GPT-6 Astra (long context)> 272K input tokens | $20.00 | $2.00 | $25.00 | $75.00 |
| GPT-6 Luna≤ 272K input tokens | $0.10 | $0.01 | $0.125 | $0.50 |
| GPT-6 Luna (long context)> 272K input tokens | $0.20 | $0.02 | $0.25 | $0.75 |
| GPT-6 Sol≤ 272K input tokens | $2.00 | $0.20 | $2.50 | $10.00 |
| GPT-6 Sol (long context)> 272K input tokens | $4.00 | $0.40 | $5.00 | $15.00 |
| GPT-6.1 Sol≤ 272K input tokens | $2.00 | $0.10 | $2.50 | $10.00 |
| GPT-6.1 Sol (long context)> 272K input tokens | $4.00 | $0.20 | $5.00 | $15.00 |
| GPT-6.1 Terra≤ 272K input tokens | $1.00 | $0.05 | $1.25 | $5.00 |
| GPT-6.1 Terra (long context)> 272K input tokens | $2.00 | $0.10 | $2.50 | $7.50 |
| Claude Haiku 4.5 | $1.00 | $0.10 | $1.25 | $5.00 |
| Claude Haiku 5.5≤ 100K input tokens | $0.10 | $0.01 | $0.125 | $0.50 |
| Claude Haiku 5.5 (long context)> 100K input tokens | $0.50 | $0.05 | $0.625 | $2.50 |
| Claude Sonnet 4 | $3.00 | $0.30 | $3.75 | $15.00 |
| Claude Sonnet 4.6 | $3.00 | $0.30 | $3.75 | $15.00 |
| Claude Opus 4.8 | $5.00 | $0.50 | $6.25 | $25.00 |
| Claude Opus 5 | $5.00 | $0.50 | $6.25 | $25.00 |
| Claude Opus 5.5 | $4.00 | $0.20 | $5.00 | $20.00 |
| Claude Sonnet 5 | $2.00 | $0.20 | $2.50 | $10.00 |
| Claude Sonnet 5.5 | $2.00 | $0.10 | $2.50 | $10.00 |
| Claude Opus 4.8 (fast mode, preview) | $10.00 | $1.00 | $12.50 | $50.00 |
| Claude Fable 5 | $10.00 | $1.00 | $12.50 | $50.00 |
| Claude Fable 5.1 | $10.00 | $0.25 | $12.50 | $50.00 |
| Gemini 3.8 Flashpromotional price through Dec 31, 2026 | $0.75 | $0.075 | — | $3.75 |
| Gemini 4 Argonpromotional price until Dec 1, 2026 | $2.00 | $0.10 | — | $10.00 |
| MAI-Code-1.1-Flash | $0.20 | $0.02 | — | $1.20 |
| Grok 4.5≤ 200K input tokens | $2.00 | $0.50 | — | $6.00 |
| Grok 4.5 (long context)> 200K input tokens | $4.00 | $1.00 | — | $12.00 |
| Grok 4.6≤ 200K input tokens | $2.00 | $0.50 | — | $6.00 |
| Grok 4.6 (long context)> 200K input tokens | $4.00 | $1.00 | — | $12.00 |
| Grok 4.7≤ 200K input tokens | $2.00 | $0.50 | — | $6.00 |
| Grok 4.7 (long context)> 200K input tokens | $4.00 | $1.00 | — | $12.00 |
| Kimi K3 | $3.00 | $0.30 | — | $15.00 |
Use real token counts, not guesses
GitHub's usage dashboard shows credits spent. For token-level detail from GitHub Copilot CLI, the whoburnedmore CLI command npx whoburnedmore reads Copilot CLI's local logs and shows its token counts per model next to Claude Code, Codex and other tools. It does not read VS Code Copilot Chat or your remaining Copilot balance. By default it signs in and syncs daily aggregates (never prompts or code); add --local to keep the report offline on your machine. Steps are in how to check GitHub Copilot usage.
Frequently asked questions
How many AI credits do I need for GitHub Copilot?
Add up your tokens per model and multiply by GitHub's per-1M-token rate, then by 100. As a rough guide from the presets above: light daily chat on Claude Sonnet 5.5 is about 34 credits a day, or about 750 a month of workdays, which fits Pro's 1,500. A typical agent-mode day on the same model is about 400 credits, or 8,800 a month, which is more than Pro+ (7,000) and fits Max (20,000). A heavy Claude Opus 5.5 agent day can pass 1,700 credits. Measure your own tokens before you pick a plan.
How do premium requests convert to AI credits?
There is no conversion rate. Premium requests counted prompts times a model multiplier. AI credits count tokens times a model's price. The only like-for-like comparison is dollars: old overage was $0.04 per premium request, which is the price of 4 AI credits. The same prompt can now cost far less (a short question on a cheap model) or far more (a long agent run that re-sends a large context many times).
Do code completions use AI credits?
No. GitHub says code completions and next edit suggestions are not billed in AI credits and stay unlimited on paid plans, even when a user has a $0 budget or hits an enforced spending limit. Copilot Chat, Copilot CLI, the cloud agent, Spaces, Spark, third-party coding agents and code review do use credits.
What happens when my AI credits run out?
On Pro, Pro+ and Max you can set an additional-usage budget (credits cost $0.01 each, so $10 covers 1,000), upgrade, or wait for the reset at 00:00 UTC on the 1st. An upgrade costs only the price difference, and the month's usage counts against the bigger allowance. On Business and Enterprise, paid usage is on by default once the pool is empty, unless an admin turns off the AI credits paid usage policy.
How does the Business and Enterprise pool work?
Each Business license adds 1,900 credits and each Enterprise license adds 3,900 to one shared pool for the billing entity, so light users offset heavy ones. Adding licenses mid-cycle grows the pool right away. Removing them only shrinks it from the next cycle. Unused credits do not roll over. Budgets cap metered charges after the pool runs out, but they never cap license fees.
Does auto model selection lower the credit cost?
Yes. On paid plans GitHub gives a 10% discount on model costs when you use auto model selection in Copilot Chat or the Copilot cloud agent. Tick the auto box on a model row to apply it. Separately, GitHub lists a 30% discount for Copilot Fusion through November 30, 2026; this calculator does not apply it.
Why might my bill differ from this estimate?
Long-context tiers are priced per request: a request above the model's input threshold (for example 272K tokens on GPT-6 Sol) uses the higher row, so put those tokens on the long-context option. Copilot code review picks its model automatically and also uses GitHub Actions minutes. Image generation is priced separately. Gemini 3.8 Flash and Gemini 4 Argon are on promotional prices that end on December 31 and December 1, 2026.
Sources checked November 8, 2026. Plan prices and allowances, per-model rates, the 10% auto model selection discount and the 400-license budget example all come from GitHub's docs listed below.
Sources
- GitHub Docs: Models and pricing for GitHub Copilot
Per-1M-token input, cached input, cache write and output rates for every model, long-context tiers, Gemini promotional end dates, and 1 AI credit = $0.01. Checked November 8, 2026.
- GitHub Docs: Usage-based billing for individuals
Pro 1,500, Pro+ 7,000 and Max 20,000 credits a month, reset at 00:00 UTC on the 1st, the 10% auto model selection discount, and additional-usage budgets. Checked November 8, 2026.
- GitHub Docs: Usage-based billing for organizations and enterprises
Business 1,900 and Enterprise 3,900 credits per user, pooled at the billing entity; paid usage on by default; completions not billed. Checked November 8, 2026.
- GitHub Docs: Budgets for usage-based billing
The example that 400 Business licenses cost $7,600 a month and a $5,000 enterprise budget allows up to $12,600 in combined charges. Checked November 8, 2026.
- GitHub Docs: Plans for GitHub Copilot
Plan prices: Pro $10, Pro+ $39, Max $100, Business $19 and Enterprise $39 per seat per month. Checked November 8, 2026.
- GitHub Docs: Requests in GitHub Copilot (legacy)
The legacy premium request allowances (300 on Pro, 1,500 on Pro+) and the $0.04 price per extra request. Checked November 8, 2026.
- GitHub Changelog: Updates to GitHub Copilot billing and plans
June 1, 2026: all Copilot plans bill on GitHub AI Credits, user-level budgets and Copilot Max launch.