Your files stay on your device
LLM Model Comparison Table: Context and Prices by Snapshot
The dated model registry as a table: context, output ceiling, rates, long tiers, tokenizer status.
registry snapshot 2026-08-26 · 33 models · no live vendor requests, ever
| Model | Family | Context | Max out | In / 1M | Cached / 1M | Out / 1M | Tokenizer |
|---|---|---|---|---|---|---|---|
| gpt-5.6-sol | GPT-5.6 | 1,050,000 | 128,000 | $4 | $0.4 | $20 | o200k_base (exact) |
| gpt-5.6-terra | GPT-5.6 | 1,050,000 | 128,000 | $2 | $0.2 | $12 | o200k_base (exact) |
| gpt-5.6-luna | GPT-5.6 | 1,050,000 | 128,000 | $0.2 | $0.02 | $1.2 | o200k_base (exact) |
| gpt-5.6-cyber | GPT-5.6 | 400,000 | 128,000 | $12.5 | $1.25 | $75 | o200k_base (exact) |
| gpt-5.5 | GPT-5.5 | 1,050,000 | 128,000 | $5 | $0.5 | $30 | o200k_base (exact) |
| gpt-5.5-pro | GPT-5.5 | 1,050,000 | 128,000 | $30 | — | $180 | o200k_base (exact) |
| gpt-5.4 | GPT-5.4 | 1,050,000 | 128,000 | $2.5 | $0.25 | $15 | o200k_base (exact) |
| gpt-5.4-mini | GPT-5.4 | 400,000 | 128,000 | $0.75 | $0.075 | $4.5 | o200k_base (exact) |
| gpt-5.4-nano | GPT-5.4 | 400,000 | 128,000 | $0.2 | $0.02 | $1.25 | o200k_base (exact) |
| gpt-5.4-pro | GPT-5.4 | 1,050,000 | 128,000 | $30 | — | $180 | o200k_base (exact) |
| gpt-5.3-codex | GPT-5.3 | 400,000 | 128,000 | $1.75 | $0.175 | $14 | o200k_base (exact) |
| gpt-5.2 | GPT-5.2 | 400,000 | 128,000 | $1.75 | $0.175 | $14 | o200k_base (exact) |
| gpt-5.2-pro | GPT-5.2 | 400,000 | 128,000 | $21 | — | $168 | o200k_base (exact) |
| gpt-5.1 | GPT-5.1 | 400,000 | 128,000 | $1.25 | $0.125 | $10 | o200k_base (exact) |
| gpt-5 | GPT-5 | 400,000 | 128,000 | $1.25 | $0.125 | $10 | o200k_base (exact) |
| gpt-5-mini | GPT-5 | 400,000 | 128,000 | $0.25 | $0.025 | $2 | o200k_base (exact) |
| gpt-5-nano | GPT-5 | 400,000 | 128,000 | $0.05 | $0.005 | $0.4 | o200k_base (exact) |
| gpt-5-pro | GPT-5 | 400,000 | 272,000 | $15 | — | $120 | o200k_base (exact) |
| gpt-4.1 | GPT-4.1 | 1,047,576 | 32,768 | $2 | $0.5 | $8 | o200k_base (exact) |
| gpt-4.1-mini | GPT-4.1 | 1,047,576 | 32,768 | $0.4 | $0.1 | $1.6 | o200k_base (exact) |
| gpt-4.1-nano | GPT-4.1 | 1,047,576 | 32,768 | $0.1 | $0.025 | $0.4 | o200k_base (exact) |
| gpt-4o | GPT-4o | 128,000 | 16,384 | $2.5 | $1.25 | $10 | o200k_base (exact) |
| gpt-4o-mini | GPT-4o | 128,000 | 16,384 | $0.15 | $0.075 | $0.6 | o200k_base (exact) |
| o3 | o-series | 200,000 | 100,000 | $2 | $0.5 | $8 | o200k_base (exact) |
| o3-mini | o-series | 200,000 | 100,000 | $1.1 | $0.55 | $4.4 | o200k_base (exact) |
| o3-pro | o-series | 200,000 | 100,000 | $20 | — | $80 | o200k_base (exact) |
| o4-mini | o-series | 200,000 | 100,000 | $1.1 | $0.275 | $4.4 | o200k_base (exact) |
| o1 | o-series | 200,000 | 100,000 | $15 | $7.5 | $60 | o200k_base (exact) |
| chat-latest | OpenAI | 400,000 | 128,000 | $5 | $0.5 | $30 | o200k_base (exact) |
| gpt-4-turbo | GPT-4 (legacy) | 128,000 | 4,096 | $10 | — | $30 | cl100k_base (exact) |
| gpt-3.5-turbo | GPT-3.5 (legacy) | 16,385 | 4,096 | $0.5 | — | $1.5 | cl100k_base (exact) |
| gpt-oss-120b | gpt-oss | 131,072 | 131,072 | — | — | — | o200k_base (exact) |
| gpt-oss-20b | gpt-oss | 131,072 | 131,072 | — | — | — | o200k_base (exact) |
- gpt-5.6-sol: Short/long-context tiers exist for this model; the vendor page does not publish the exact boundary; estimates split at 272K, the documented boundary for these families, and this estimate uses short-tier (standard) prices.
- gpt-5.6-terra: Short/long-context tiers exist for this model; the vendor page does not publish the exact boundary; estimates split at 272K, the documented boundary for these families, and this estimate uses short-tier (standard) prices.
- gpt-5.6-luna: Short/long-context tiers exist for this model; the vendor page does not publish the exact boundary; estimates split at 272K, the documented boundary for these families, and this estimate uses short-tier (standard) prices.
- chat-latest: Rolling alias: the underlying model and price change when OpenAI moves it.
- gpt-oss-120b: Open weights: API pricing depends on whoever hosts it; no single price is honest here.
- gpt-oss-20b: Open weights: API pricing depends on whoever hosts it; no single price is honest here.
- gpt-5.6-sol long tier: > 272K in: $8 / $30
- gpt-5.6-terra long tier: > 272K in: $4 / $18
- gpt-5.6-luna long tier: > 272K in: $0.4 / $1.8
- gpt-5.5 long tier: > 272K in: $10 / $45
- gpt-5.5-pro long tier: > 272K in: $60 / $270
- gpt-5.4 long tier: > 272K in: $5 / $22.5
- gpt-5.4-pro long tier: > 272K in: $60 / $270
| Field | gpt-5-nano | gpt-5-pro |
|---|---|---|
| Context window | 400,000 | 400,000 |
| Max output | 128,000 | 272,000 |
| Input / 1M | $0.05 | $15 |
| Cached input / 1M | $0.005 | — |
| Output / 1M | $0.4 | $120 |
| Encoding | o200k_base | o200k_base |
Context delta +0 tokens · max-output delta −144,000 tokens (left − right).
- Every number comes from the dated registry snapshot (2026-08-26), prices curated from the vendor pricing page that day, context windows cross-checked against the installed tokenizer package. Pricing curated 2026-08-26 from developers.openai.com/api/docs/pricing (standard tier, USD per 1M tokens); context windows and max output cross-checked against the installed gpt-tokenizer 4.0.0 model catalog. Prices change; re-check the vendor page before relying on an estimate.
- Tokenizer status: every model in this registry has a shipped local encoding, so counts for them are exact. The "estimate" path other tools carry for non-registry vendors never appears here.
- This page makes no live vendor requests; it cannot check today’s price for you. Re-verify against the vendor page before relying on a number.
Done
Finished here in the tab. Nothing was uploaded.
snapshot 2026-08-26: 33 rows shown