Trending on Kingy
Keep reading with the stories getting the most attention now.
Verdict: DeepSeek V4 Pro 0813 is live as an API model, and DeepSeek’s own launch table puts it remarkably close to Claude Fable 5 on a few agent benchmarks. But “Fable-level at 1/57 the price” compresses three different claims into one slogan. DeepSeek’s comparison chart is vendor-reported. A new independent run from Artificial Analysis supports a substantial improvement over April’s V4 Pro, but it does not establish broad Fable 5 equivalence. The 57-times price ratio is true only for output tokens, and the official DeepSeek repository still does not establish that its downloadable weights are the 0813 build.
This is a breaking-news source audit, not a definitive hands-on review. Kingy did not run controlled prompts, measure latency, inspect API outputs, or reproduce the benchmark harnesses. What can be verified today is narrower: the model is callable, the published price is unusually low, and an independent composite evaluation now supports a large gain over April’s V4 Pro. What cannot yet be verified is broad Fable 5 equivalence, reproduction of DeepSeek’s selected agent chart, or an 0813 open-weight release.
Update, August 13 UTC: Artificial Analysis has published the first independent 0813 result found in this follow-up check. Its current Intelligence Index score rises from 45 for April’s V4 Pro to 53 for 0813. DeepSeek’s official changelog still ends with the July 31 V4 Flash update, and the official V4 Pro repository remains last modified June 22 with no 0813 tag or revision. The new independent result strengthens the case that 0813 is a real upgrade; it does not validate every row in DeepSeek’s launch chart or resolve the weight-checkpoint question.
| Claim | Fact-check | Why |
|---|---|---|
| “0813 is live” | Supported for API access | DeepSeek’s pricing page names DeepSeek-V4-Pro-0813; OpenRouter lists the 0813 model as the GA release. |
| “Fable-level” | Partly supported, too broad as a blanket claim | DeepSeek reports near-ties on Terminal Bench 2.1 and CyberGym, but larger gaps on HLE, DeepSWE, Toolathlon and DSBench-FullStack. |
| “1/57 the price” | True for output only | $0.87 versus $50 per million output tokens. Base input is about 1/23; a simple equal input/output blend is about 1/46. |
| “0813 weights are available” | Unconfirmed | The official V4 Pro repository has downloadable weights, but no August commit, 0813 tag or revision identifying those files as the new API build. |
What Is Actually Live?
DeepSeek’s official Models & Pricing page now identifies the current deepseek-v4-pro version as DeepSeek-V4-Pro-0813. The API alias stays the same. The same page lists a one-million-token context window, a maximum output of 384,000 tokens, thinking and non-thinking modes, tool calls, JSON output, the Responses API and Anthropic-compatible access. Artificial Analysis and OpenRouter both list the route as text-to-text; neither record advertises image input.
OpenRouter’s model record independently confirms a live route under deepseek/deepseek-v4-pro-0813, served by DeepSeek, and describes it as “the GA release of DeepSeek V4 Pro.” OpenRouter’s launch post says the model is live and attributes the reported agent gains to DeepSeek.
There is one reason to keep “silent launch” in quotation marks. DeepSeek’s official changelog has no August 13 entry at publication time. Its latest entry is the July 31 V4 Flash update, which said the official V4 Pro release would follow soon. The live pricing record and OpenRouter route are sufficient to establish API availability; the missing changelog entry makes the rollout unusually under-documented, not imaginary.
What the Fuller Official Specs Add
DeepSeek’s official repository describes V4 Pro as a mixture-of-experts model with 1.6 trillion total parameters and 49 billion activated per token, trained as part of the V4 family on more than 32 trillion tokens. At a one-million-token context, DeepSeek says the architecture needs 27% of V3.2’s single-token inference FLOPs and 10% of its KV-cache footprint. These are DeepSeek-reported architecture and training statistics for the V4 release family; the repository still does not identify its files or model-card evaluations as the 0813 API revision.
The current thinking-mode documentation also makes benchmark configuration more consequential than a single launch score suggests. Thinking is enabled by default at high effort. Requests for medium, high or xhigh all map to the same high setting, while max selects a separate maximum-effort mode. In thinking mode, temperature, top_p, presence penalty and frequency penalty are accepted for compatibility but have no effect. DeepSeek’s rate-limit documentation lists a per-account concurrency limit of 500 for V4 Pro.
OpenRouter adds one useful live-serving detail: its model record says 0813 is currently hosted by one provider and requests are forwarded directly to DeepSeek. Its public endpoint record currently leaves latency and throughput fields empty. That absence is not evidence of slow service; it means OpenRouter does not yet provide a durable serving measurement worth citing.
Reasoning Effort Can Move the Reported Scores Sharply
The official V4 model card publishes mode-by-mode results for the earlier V4 Pro release family. They are material because they show how strongly scores can depend on reasoning budget. They are not 0813 evaluations, and they use different benchmark versions and harnesses from the new agent chart, so the two tables should not be merged.
| DeepSeek-reported repository eval | V4 Pro non-think | V4 Pro high | V4 Pro max | Non-think to max |
|---|---|---|---|---|
| LiveCodeBench (Pass@1) | 56.8 | 89.8 | 93.5 | +36.7 points |
| HLE (Pass@1) | 7.7 | 34.5 | 37.7 | +30.0 points |
| MRCR 1M (MMR) | 44.7 | 83.3 | 83.5 | +38.8 points |
| SWE-bench Verified (resolved) | 73.6 | 79.4 | 80.6 | +7.0 points |
The practical conclusion is narrower than “max is always better.” In DeepSeek’s own table, moving from high to max adds 3.7 points on LiveCodeBench and 3.2 on HLE, but only 0.2 on MRCR 1M and 1.2 on SWE-bench Verified. Evaluation claims therefore need the model revision, effort setting, harness, benchmark version and token budget. The 0813 launch graphic supplies scores but not enough of that methodology for independent reproduction.
The First Independent Result Has Arrived
Artificial Analysis now has a dedicated V4 Pro 0813 page and describes its measurement as an independent run on dedicated hardware. At max reasoning effort, 0813 scores 53 on Artificial Analysis Intelligence Index v4.1.1, versus 45 for the April V4 Pro baseline. The current index combines nine evaluations, including Terminal-Bench 2.1, Humanity’s Last Exam, GPQA Diamond, SciCode, GDPval-AA v2 and long-context and agentic components.
| Independent Artificial Analysis measure | V4 Pro April | V4 Pro 0813 | Change |
|---|---|---|---|
| Intelligence Index | 45 | 53 | +8 points |
| Output speed | 70.6 tokens/s | 83.2 tokens/s | About 18% faster |
| Time to first token | 1.68s | 1.63s | 0.05s faster |
| Output tokens used by the index | 180M | 130M | About 28% fewer |
| Total index evaluation cost | $176.25 | $135.03 | About 23% lower |
This is the most important post-publication addition. It independently supports the claim that 0813 is materially stronger and more token-efficient than the April route. It still does not reproduce DeepSeek’s Fable comparison: Artificial Analysis uses its own composite, settings and graders, and the public model page does not turn a score of 53 into a blanket Fable 5 equivalence claim.
One metadata caveat remains. Artificial Analysis labels 0813 as open weights, but its page does not identify an 0813-specific repository revision or checksum. The direct DeepSeek repository check remains the more precise evidence for checkpoint availability: the public files exist for the V4 Pro family, while their correspondence to the API-served 0813 build is still unconfirmed.
The Benchmark Chart Is DeepSeek-Reported
The supplied benchmark image carries DeepSeek branding and compares V4 Pro 0813 with V4 Flash 0731, the two V4 previews, GLM-5.2, Kimi-K3, Opus 4.8 and “Fable 5 (w/ fallback).” Those numbers should be read as DeepSeek-reported launch results, not Kingy measurements and not an independent leaderboard.
| Benchmark | DeepSeek V4 Pro 0813 | Fable 5 (with fallback) | Difference |
|---|---|---|---|
| HLE without / with tools | 42.7 / 60.0 | 53.3 / 63.0 | DeepSeek trails by 10.6 / 3.0 points |
| Terminal Bench 2.1 | 87.9 | 88.0 | DeepSeek trails by 0.1 point |
| CyberGym | 83.3 | 83.1 | DeepSeek leads by 0.2 point |
| DeepSWE | 62.7 | 70.0 | DeepSeek trails by 7.3 points |
| Toolathlon-Verified | 74.1 | 77.9 | DeepSeek trails by 3.8 points |
| AutomationBench (Public) | 31.8 | 29.1 | DeepSeek leads by 2.7 points |
| DSBench-FullStack | 71.1 | 77.2 | DeepSeek trails by 6.1 points |
| DSBench-Hard | 67.2 | 68.3 | DeepSeek trails by 1.1 points |
The honest reading is Fable-adjacent on parts of this selected agent suite. DeepSeek is effectively tied on Terminal Bench 2.1 and CyberGym, leads on AutomationBench, and comes within 1.1 points on DSBench-Hard. It is not tied on every row: Fable leads by 7.3 points on DeepSWE, 6.1 on DSBench-FullStack and 10.6 on HLE without tools.
The fallback label also matters. Anthropic’s official Fable 5 documentation says queries flagged by its cybersecurity and biology safeguards may be routed to less capable models. DeepSeek’s comparison therefore describes Anthropic’s deployed Fable-with-fallback product behavior on this chart; it does not establish equivalence to every possible Fable configuration or workload.
The Percentage-Point Correction
Fresh X posts repeated the preview-to-0813 jumps, sometimes presenting the scores as percentages. The subtraction is measured in percentage points when both values are rates on a 0–100 scale. Relative percentage growth is a different calculation.
| DeepSeek-reported benchmark | V4 Pro Preview | V4 Pro 0813 | Correct point gain | Relative increase |
|---|---|---|---|---|
| Terminal Bench 2.1 | 72.1 | 87.9 | +15.8 points | +21.9% |
| CyberGym | 52.7 | 83.3 | +30.6 points | +58.1% |
| DeepSWE | 12.8 | 62.7 | +49.9 points | +389.8% |
| AutomationBench (Public) | 12.8 | 31.8 | +19.0 points | +148.4% |
| DSBench-Hard | 31.1 | 67.2 | +36.1 points | +116.1% |
Calling DeepSWE a “49.9% improvement” would understate the relative change and misuse the unit. The score rose 49.9 percentage points; relative to the 12.8 starting value, the increase is about 390%. Neither number proves real-world task completion without the harness, settings, sample sizes and reproducibility data.
Is It Really 1/57 the Price of Fable 5?
Only if the comparison is output tokens. DeepSeek charges $0.435 per million cache-miss input tokens, $0.003625 per million cache-hit input tokens and $0.87 per million output tokens. Anthropic lists Fable 5 at $10 per million base input tokens and $50 per million output tokens.
| Billing component | DeepSeek V4 Pro 0813 | Claude Fable 5 | DeepSeek share of Fable price |
|---|---|---|---|
| Base / cache-miss input | $0.435 / 1M | $10 / 1M | About 1/23 |
| Cache read / hit input | $0.003625 / 1M | $1 / 1M | About 1/276 |
| Output | $0.87 / 1M | $50 / 1M | About 1/57 |
| Simple 1M input + 1M output example | $1.305 | $60 | About 1/46 |
Real agent bills depend on the input/output mix, reasoning-token treatment, cache hits, retries and tool-loop length. Fable’s prompt cache writes also have their own rates: Anthropic lists $12.50 per million tokens for a five-minute cache write and $20 for a one-hour write, then $1 for cache reads. DeepSeek warns that its current overall API pricing will rise significantly in the near future, with the exact plan still subject to a later notice. The launch-day rates are verified, not guaranteed indefinitely.
KV Cache Is Not the Same as Prompt Cache Pricing
The official DeepSeek V4 repository says V4 Pro requires 10% of DeepSeek V3.2’s KV-cache footprint at a one-million-token context. That is an inference-memory efficiency claim: the key-value cache stores attention state while a model processes and generates from a long sequence.
DeepSeek’s API context-caching feature is a separate customer-facing mechanism. It stores reusable prompt prefixes on disk, is enabled by default, and bills matching input tokens at the cache-hit rate. DeepSeek calls it best-effort and does not guarantee every repeated prefix will hit.
Anthropic’s prompt caching likewise reuses prompt prefixes, but exposes explicit or automatic cache controls, five-minute or one-hour lifetimes, and separate write/read prices. These concepts are related inside inference systems, but the numbers are not interchangeable. “90% less KV cache” does not mean “90% off the API bill,” and a DeepSeek cache-hit price should not be compared with a Fable base-input price as if every request automatically qualifies on both sides.
Are the 0813 Open Weights Available?
Not confirmed. DeepSeek’s official V4 Pro repository does contain MIT-licensed model weights for the V4 Pro release family. But its commit history shows no August update: the latest commit is a June 22 technical-report change, while the large weight files arrived with the April preview release. The model card still introduces a “preview version” of the V4 series and does not name 0813.
That evidence cannot support the X claim that users already have access to the 0813 weights. It supports a narrower statement: V4 Pro weights exist, while correspondence between those repository files and the new API-served 0813 checkpoint is unresolved. Confirmation would require an explicit DeepSeek revision, tag, commit, checksum change or model-card statement.
What the Audited X Posts Got Right—and What They Overstated
- ChrisGPT correctly attributed the benchmark jumps to DeepSeek and quoted the live price and context window. The post also noted the problem of companies reporting different benchmark sets.
- DanDr1s accurately reproduced five rows from the supplied DeepSeek chart and the $0.435/$0.87 rates. The gains should be described as points, not percentages.
- Junyuan Shang called the result “nearly Fable 5-level.” That is defensible for a subset of the chart, but too broad without the rows where Fable leads and without independent replication.
- Chizzydigital paired the Fable-level framing with “access to the weights.” Generic V4 Pro weights are public; the 0813 weights are not yet identified as such by DeepSeek.
Kingy’s Breaking-News Read
DeepSeek V4 Pro 0813 is one of the most aggressive price-performance launches of 2026. Artificial Analysis’s independent score of 53—up from 45 for April V4 Pro—now supports a real capability gain, while its lower token use and lower evaluation cost strengthen the price-performance case. DeepSeek’s near-parity claim with Fable 5 on Terminal Bench 2.1 and CyberGym remains newsworthy but vendor-reported.
But the evidence does not justify a universal “Fable for 1/57 the cost” verdict. DeepSeek’s chart is vendor-reported, its Fable column is explicitly a fallback configuration, the models trade wins across the selected rows, and the 57-times figure applies only to output. The safest current conclusion is: 0813 is a live, exceptionally inexpensive text-only API model with an independently measured improvement over April V4 Pro and vendor-reported Fable-adjacent agent performance. Reproduction of DeepSeek’s launch chart and an explicit 0813 weight release are still pending.
FAQ
Is DeepSeek V4 Pro 0813 live?
Yes for API access. DeepSeek’s official pricing page identifies the current deepseek-v4-pro version as DeepSeek-V4-Pro-0813, and OpenRouter has a live 0813 route.
Does DeepSeek V4 Pro 0813 match Claude Fable 5?
DeepSeek reports near-ties on some agent benchmarks, including Terminal Bench 2.1 and CyberGym. Fable leads by wider margins on other rows such as DeepSWE, HLE without tools and DSBench-FullStack. No independent Kingy test has established broad equivalence.
Is DeepSeek really 57 times cheaper than Fable 5?
For output tokens, yes: $0.87 versus $50 per million is about 57.5 times cheaper. Base input is about 23 times cheaper, and the total ratio depends on each workload’s token mix and cache behavior.
Are the DeepSeek V4 Pro 0813 weights open?
Unconfirmed. The official repository has V4 Pro weights, but it does not currently identify them as the 0813 checkpoint and has no August commit or 0813 tag.
Does DeepSeek V4 Pro 0813 support images?
No image input is listed in the current Artificial Analysis or OpenRouter records; both describe 0813 as text-to-text. Treat the API model as text-only unless DeepSeek documents a vision capability.
What is the difference between KV cache and prompt caching?
KV cache is the model’s attention-state memory during inference. Prompt caching is an API feature that reuses matching input prefixes across requests and can change billed input cost. A reduction in KV-cache memory does not directly equal the prompt-cache discount.
Sources
- DeepSeek API: Models & Pricing
- DeepSeek API: Thinking Mode and effort mapping
- DeepSeek API: Rate Limit & Isolation
- DeepSeek API changelog
- DeepSeek API: Context Caching
- Official DeepSeek V4 Pro repository and model card
- Official DeepSeek V4 Pro commit history
- OpenRouter model record: DeepSeek V4 Pro 0813
- OpenRouter launch post on X
- Anthropic: Claude Fable 5
- Anthropic: Prompt caching and Fable 5 cache pricing
- Artificial Analysis: DeepSeek V4 Pro 0813 independent evaluation
- Artificial Analysis: April V4 Pro baseline
- Artificial Analysis Intelligence Index methodology
- Audited X post: ChrisGPT
- Audited X post: DanDr1s
- Audited X post: Junyuan Shang
- Audited X post: Chizzydigital