DeepSeek V4 Pro 0813 vs GPT-5.6 Sol
Official off-peak $0.66/$1.98 vs $4/$20, context 1M vs 1,050,000
Sol is GPT-5.6's top tier (GA 2026-07-09) with a 1,050,000 context window. Pro 0813 is DeepSeek's current flagship (GA 2026-08-13) with 1M context, 384K max output and a concurrency limit of 500. Prompts over 272K input tokens are priced at 2x input and 1.5x output for the full request, and promotional pricing runs until 2026-11-30. Checked 2026-10-20.
Updated 2026-11-10
Switch models on the same endpoint and measure the cost per token yourself; check /models for callable models.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Highlights
DeepSeek side
Official pricing page, checked 2026-10-20: model version DeepSeek-V4-Pro-0813, 1M context, 384K max output, concurrency limit 500.
OpenAI top tier
Sol scores 47 on the Artificial Analysis Intelligence Index (Max, third-party index checked 2026-10-20). Official now lists $4/M in / $20/M out (promo until 2026-11-30, 2026-10-20 check of the official page); the official page states the current price against the launch $5/$30 is a 20% cut in input pricing and a 33% cut in output pricing.
Third-party index, 0813
The third-party index (Artificial Analysis, community-reported, fetched 2026-10-20) scores DeepSeek V4 Pro 0813 at 36; such figures move with the index version, so check that page when you need them.
DeepSeek pricing
Official prices for deepseek-v4-pro per 1M tokens: cache hit $0.022 (peak $0.044), cache miss $0.66 (peak $1.32), output $1.98 (peak $3.96).
The matchup
Searchers want a single winner. Give them a job split: Sol for polished long reasoning and the OpenAI product surface; 0813 for open agents, and for lower token bills when you call DeepSeek's official API directly. The official prices, context windows and long-context billing rules of both are in the highlights and comparisons above (checked 2026-10-20); QCode rates follow the live prices on qcode.cc/models.
Why it is trending
GPT-5.6 has Sol/Terra/Luna. This page's opponent is Sol only. The existing Sol vs Terra vs Luna guide is at /gpt-5-6-sol-vs-terra-vs-luna.
Timeline
GPT-5.6 GA, Sol is the top tier.
Pro 0813 GA.
DeepSeek official pricing moves to peak/off-peak rates (from 16:00 UTC).
Confirmed vs caution
Usable now
Official DeepSeek 0813 pricing page and docs (checked 2026-10-20): deepseek-v4-pro is model version DeepSeek-V4-Pro-0813 with 1M context, 384K max output and a concurrency limit of 500; deepseek-v4-flash is served by DeepSeek-V4.1-Flash-1017, and the legacy names deepseek-v4-flash and deepseek-v4-flash-vision-exp are still accepted although those models are retired. Sol's public AA 47 comes from the third-party index (community-reported, fetched 2026-10-20), and current pricing is $4/$20 (promotional until 2026-11-30); prompts over 272K input tokens are priced at 2x input and 1.5x output for the full request.
Conclusions that do not hold
Pre-0813 AA 45 as if it were this build. Third-party $3.75/$22.5 Sol quotes without OpenAI confirmation.
Pick by job
Pick Sol
Closed flagship reasoning, Codex-native stories, with an official context window of 1,050,000 and the long-context billing rule above 272K input tokens.
Pick 0813
Self-host, open agents, huge volume, with the official pricing page listing off-peak and peak official prices, 1M context, 384K max output and a concurrency limit of 500.
How to use it
To choose within GPT-5.6, see /gpt-5-6-sol-vs-terra-vs-luna. For V4 Pro 0813's full specs and benchmarks, see /deepseek-v4-pro-0813. The official DeepSeek pricing page lists two BASE URLs: OpenAI format https://api.deepseek.com and Anthropic format https://api.deepseek.com/anthropic (checked 2026-10-20).
On QCode
QCode currently lists both deepseek-v4-pro and gpt-5.6-sol on /models, so you can compare them with a single key; deepseek-v4.1-flash is listed as well. Billing is per token and per-model rates are on /models. Want to try a new model for a few days? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 days; if you already have a Personal Plan and run short, buy a booster pack.
FAQ
How do I choose between DeepSeek V4 Pro 0813 and GPT-5.6 Sol?
On the Artificial Analysis intelligence index (third-party index, community-reported, fetched 2026-10-20), GPT-5.6 Sol scores 47 and DeepSeek V4 Pro 0813 scores 36. That is community data rather than an official conclusion; the official prices, context windows and long-context billing rules are listed above (checked 2026-10-20).
What is the current AA index?
Artificial Analysis index as fetched on 2026-10-20: Sol (Max) 47, V4 Pro 0813 (Max) 36. These replace the earlier Sol 61 and 45/53 figures, which should no longer be quoted.
Context windows?
Pro 1M official. Sol often listed ~1.05M API — confirm on OpenAI/QCode models.
Codex?
DeepSeek added native Responses/Codex notes on Flash and Pro. Sol is native to the OpenAI/Codex world.
Terra/Luna instead?
GPT-5.6 Terra and Luna are cheaper tiers with their own pricing and capability profile — see /gpt-5-6-sol-vs-terra-vs-luna.
How are DeepSeek's official prices calculated (off-peak and peak)?
The official page states that off-peak rates are half of the peak rates; peak hours are 01:00–04:00 and 06:00–10:00 UTC, Monday through Friday, excluding Chinese public holidays, and all other hours are off-peak, including weekends and Chinese public holidays in full.
Sources
DeepSeek official Models & Pricing page and official docs; OpenAI GPT-5.6 model page and pricing page; the third-party index Artificial Analysis model pages (community-reported, fetched 2026-10-20) give Sol 47 and V4 Pro 0813 36; the model ids sold on QCode /models. All checked on 2026-10-20.
Start from the two official rate cards
Engineers can flip one paragraph when the official pages change. Want to try a new model for a few days without a monthly plan? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 days; with a Personal Plan, buy a booster pack.
Related
0813 guide
GA specs.
Pro vs Fable 5
Same-table agent grid.
Sol vs Terra vs Luna
Existing GPT-5.6 cluster.
Crypto payment guide
Nine coins, 56 networks, zero platform fee for AI coding API credits.
Not affiliated with DeepSeek or OpenAI. Checked on 2026-10-20 against the official DeepSeek and OpenAI pages; model availability follows /models.