# aisubsidy: full catalogue

> Every OAuth coding plan, what it costs, and the most it could possibly hand you.

Generated from the source data. 7 vendors, 19 plans, 14 models.

## Alibaba (Qwen Code)

Authentication: oauth. Checked 2026-08-20. Homepage: https://qwenlm.github.io/qwen-code-docs/en/users/configuration/auth/

### API rates

- Qwen3.8 Max (qwen3.8-max): $2.00 per million input tokens, $6.00 per million output tokens. cached input $0.25
- Qwen3 Coder Next (qwen3-coder-next): no first-party per-token rate sourced. served on the Coding Plan; standalone API rate not verified first-party

### Plans

#### Coding Plan (qwen-coding-plan)

Price: $50 per month.
The plan allows 6,000 requests per 5 hours and 90,000 requests per month.
The free OAuth tier (1,000/day, later 100/day) and the 2,000/day Qwen Code CLI tier both closed 2026-04-15.
Quota type: request_count. Basis: published.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: qwen3-coder-next, qwen3-coder-plus, qwen3-max, kimi-k2.5, glm-5, minimax-m2.5.
Sources: https://qwenlm.github.io/qwen-code-docs/en/users/configuration/auth/

## Anthropic (Claude Code)

Authentication: oauth. Checked 2026-08-20. Homepage: https://claude.com/product/claude-code

### API rates

- Claude Opus 5 (claude-opus-5): $5.00 per million input tokens, $25.00 per million output tokens.
- Claude Fable 5 (claude-fable-5): $10.00 per million input tokens, $50.00 per million output tokens.
- Claude Sonnet 5 (claude-sonnet-5): $3.00 per million input tokens, $15.00 per million output tokens. intro $2/$10 through 2026-08-31
- Claude Haiku 4.5 (claude-haiku-4-5): $1.00 per million input tokens, $5.00 per million output tokens.

### Plans

#### Pro (claude-pro)

Price: $20 per month.
The plan allows the baseline allowance per 5 hours and an unpublished cap per week.
Anthropic publishes multipliers, not token counts. Only measurement resolves this.
Quota type: rate_window. Basis: unknown.
Maximum possible spend at list API rates: about $15,581 per month (claude-max-5x measured, scaled by the published 1x multiplier). This is a ceiling assuming every window runs flat out, not a forecast.
Models: claude-opus-5, claude-sonnet-5, claude-haiku-4-5.
Sources: https://claude.com/pricing

#### Max 5x (claude-max-5x)

Price: $100 per month.
The plan allows 5 times the allowance of Pro per 5 hours and an unpublished cap per week.
Quota type: rate_window. Basis: unknown.
Maximum possible spend at list API rates: about $77,904 per month (p90 of 74 measured 5h windows, saturated across a month). This is a ceiling assuming every window runs flat out, not a forecast.
Models: claude-opus-5, claude-sonnet-5, claude-haiku-4-5.
Sources: https://claude.com/pricing

#### Max 20x (claude-max-20x)

Price: $200 per month.
The plan allows 20 times the allowance of Pro per 5 hours and an unpublished cap per week.
Quota type: rate_window. Basis: unknown.
Maximum possible spend at list API rates: about $311,616 per month (claude-max-5x measured, scaled by the published 20x multiplier). This is a ceiling assuming every window runs flat out, not a forecast.
Models: claude-opus-5, claude-sonnet-5, claude-haiku-4-5.
Sources: https://claude.com/pricing

### Measurement

Plan claude-max-5x, 30 days, 78,408 turns across 74 active 5h windows. Median $134 per window, p90 $541, max $900. Method: local Claude Code session logs bucketed into fixed 5h windows, priced at the list API rates above. Plan read from ~/.claude.json oauthAccount.organizationRateLimitTier, not assumed from the size of the spend.. One heavy user, one month, on Max 5x. No rate limit event appears in the logs for this window, so this is very likely a floor rather than a ceiling: it is what one user drew, not what the plan would have allowed. p90 is quoted rather than max. Reproduce with `node probe/windows.js`.

## MiniMax (MiniMax CLI)

Authentication: oauth. Checked 2026-08-20. Homepage: https://platform.minimax.io/docs/token-plan/minimax-cli

### API rates

- MiniMax M3 (minimax-m3): no first-party per-token rate sourced. no first-party per-token rate verified. Third-party figures for the MiniMax API start around $0.12/M but were not traceable to a MiniMax page.

### Plans

#### Token Plan Plus (minimax-plus)

Price: $20 per month.
No quota is published for this plan.
OAuth is RFC 8628 device grant with PKCE via `mmx auth login`; tokens land in ~/.mmx/config.json and refresh automatically. Token allowance per tier not verified first-party.
Quota type: token_pool. Basis: unknown.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: minimax-m3.
Sources: https://platform.minimax.io/docs/token-plan/minimax-cli

#### Token Plan Max (minimax-max)

Price: $50 per month.
No quota is published for this plan.
Quota type: token_pool. Basis: unknown.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: minimax-m3.
Sources: https://platform.minimax.io/docs/token-plan/minimax-cli

#### Token Plan Ultra (minimax-ultra)

Price: $120 per month.
No quota is published for this plan.
Quota type: token_pool. Basis: unknown.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: minimax-m3.
Sources: https://platform.minimax.io/docs/token-plan/minimax-cli

## Moonshot AI (Kimi Code)

Authentication: oauth. Checked 2026-08-20. Homepage: https://platform.moonshot.ai

### API rates

- Kimi K3 (kimi-k3): $3.00 per million input tokens, $15.00 per million output tokens. cache-hit input $0.30. 2.8T parameters, released 2026-07-17.
- Kimi K2.7 Code (kimi-k2.7-code): $0.95 per million input tokens, $4.00 per million output tokens.

### Plans

#### Moderato (kimi-moderato)

Price: $19 per month.
The plan allows 300 calls per 5 hours.
Kimi Code subscribers get their own keys with a call quota per rolling 5h window, not a token balance. Tiers span roughly 300-1200 calls. Consumer membership is priced in CNY on some surfaces; USD figures vary by source.
Quota type: request_count. Basis: derived.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: kimi-k3, kimi-k2.7-code.
Sources: https://platform.moonshot.ai

#### Allegretto (kimi-allegretto)

Price: $39 per month.
The plan allows 600 calls per 5 hours.
Quota type: request_count. Basis: derived.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: kimi-k3, kimi-k2.7-code.
Sources: https://platform.moonshot.ai

#### Allegro (kimi-allegro)

Price: $99 per month.
The plan allows 1,200 calls per 5 hours.
1M-token conversations
Quota type: request_count. Basis: derived.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: kimi-k3, kimi-k2.7-code.
Sources: https://platform.moonshot.ai

## OpenAI (Codex)

Authentication: oauth. Checked 2026-08-20. Homepage: https://developers.openai.com/codex/pricing

### API rates

- GPT-5.6 Sol (gpt-5.6-sol): $5.00 per million input tokens, $30.00 per million output tokens. after the 2026-07-30 price cut
- GPT-5.6 Terra (gpt-5.6-terra): $2.00 per million input tokens, $12.00 per million output tokens.
- GPT-5.6 Luna (gpt-5.6-luna): $0.20 per million input tokens, $1.20 per million output tokens.

### Plans

#### ChatGPT Go (codex-go)

Price: $8 per month.
The plan allows an unpublished number of messages per week.
CLI, IDE and web Codex draw from one shared message allowance
Quota type: rate_window. Basis: unknown.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: gpt-5.6-terra, gpt-5.6-luna.
Sources: https://chatgpt.com/codex/pricing/

#### ChatGPT Plus (codex-plus)

Price: $20 per month.
The plan allows an unpublished number of messages per week.
5-hour cap lifted for Plus/Business/Pro on 2026-07-12 with no published end date; the weekly cap stands
Quota type: rate_window. Basis: unknown.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna.
Sources: https://developers.openai.com/codex/pricing

#### ChatGPT Pro (codex-pro)

Price: $100 per month.
The plan allows an unpublished number of messages per week.
Quota type: rate_window. Basis: unknown.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: gpt-5.6-sol, gpt-5.6-terra, gpt-5.6-luna.
Sources: https://developers.openai.com/codex/pricing

## xAI (Grok Build)

Authentication: oauth. Checked 2026-08-20. Homepage: https://x.ai/news/grok-build-cli

### API rates

- Grok 4.6 (grok-4.6): $2.00 per million input tokens, $6.00 per million output tokens. cached input $0.50. A prompt at or above 200K tokens reprices the whole request to $4 in / $12 out. Priority Processing is 2x.

### Plans

#### SuperGrok (supergrok)

Price: $30 per month.
The plan allows an unpublished cap per week.
No standalone Grok Build subscription. Access rides an xAI or X plan.
Quota type: rate_window. Basis: unknown.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: grok-4.6.
Sources: https://x.ai/news/grok-build-cli

#### X Premium+ (x-premium-plus)

Price: $40 per month.
The plan allows an unpublished cap per week.
Quota type: rate_window. Basis: unknown.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: grok-4.6.
Sources: https://x.ai/news/grok-build-cli

#### SuperGrok Heavy (supergrok-heavy)

Price: $300 per month.
The plan allows an unpublished cap per week.
Highest limits. Grok Build launched Heavy-only on 2026-05-14 and opened to the cheaper tiers ten days later.
Quota type: rate_window. Basis: unknown.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: grok-4.6.
Sources: https://x.ai/news/grok-build-cli

## Z.ai (GLM Coding Plan)

Authentication: oauth. Checked 2026-08-20. Homepage: https://docs.z.ai/devpack/overview

### API rates

- GLM-5.3 (glm-5.3): $1.40 per million input tokens, $4.40 per million output tokens.

### Plans

#### Coding Plan Lite (glm-lite)

Price: $18 per month. $13 per month billed annually.
The plan allows 80 prompts per 5 hours, 400 prompts per week and 10,000 credits per week.
One user prompt fans out to roughly 15-20 internal model calls but counts as one against quota. Third-party figures for Pro and Max conflict ($72/$80 and $160/$168).
Quota type: request_count. Basis: published.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: glm-5.3.
Sources: https://docs.z.ai/devpack/overview

#### Coding Plan Pro (glm-pro)

Price: $80 per month. $56 per month billed annually.
The plan allows 6 times the allowance of Coding Plan Lite per week.
Quota type: request_count. Basis: published.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: glm-5.3.
Sources: https://docs.z.ai/devpack/overview

#### Coding Plan Max (glm-max)

Price: $168 per month. $118 per month billed annually.
The plan allows 14 times the allowance of Coding Plan Lite per week.
Quota type: request_count. Basis: published.
Maximum possible spend: not measured. The vendor publishes no token figure.
Models: glm-5.3.
Sources: https://docs.z.ai/devpack/overview