DeepSeek V4 Is Live on InstantClaw — What Changed and Why It Matters

DeepSeek released V4 in two flavors. The one that matters most for InstantClaw users is V4-Flash: 284B total parameters, 13B active per token, built on a Mixture-of-Experts architecture. It's not a trimmed-down version of the big model — it was trained separately, and for most everyday tasks, the gap between Flash and the full Pro variant is surprisingly narrow.

Headline improvements (see DeepSeek's model card and tech report for the full tables):

• 1M token context window—roughly 15–20 novels of text. You can put a full codebase, a long legal file, or months of chat in one session when your stack allows it.
• Think Max: DeepSeek recommends at least 384K tokens of context for the strongest reasoning mode so long, multi-step work fits. That is a context budget recommendation, not a separate “output cap” product spec—check their docs for the mode you use.
• Non-think and two thinking effort levels (fast answer vs deeper reasoning, plus Think Max for the heaviest work).
• Strong instruct scores: in DeepSeek's own tables, V4-Flash in Think Max reaches 86.2 on MMLU-Pro and 91.6 on LiveCodeBench. Other modes score lower, so the headline numbers are mode-specific.

In practice, long context used to get expensive fast. V4 is built to make million-token work far more efficient than the naive approach. You still pay for what you use; you are just on a much better cost curve for very large prompts.

Both models were pre-trained on more than 32 trillion tokens; released weights use mixed FP4 and FP8 as described on Hugging Face. Open weights are on Hugging Face under the MIT License (same family as V3; read each repo's LICENSE for your use case).

DeepSeek published a full benchmark grid against Kimi K2.6, Claude Opus 4.6, GPT-5.4, and Gemini 3.1 Pro. The honest reading: V4-Pro wins some, loses some.

Where V4 wins:

• Codeforces: 3206 rating — beats GPT-5.4 (3168) and establishes V4 as the best open-weight model for competitive programming
• LiveCodeBench: 93.5 vs K2.6's 89.6 — short-form code generation is a clear strength
• Chinese-SimpleQA: 84.4 vs the next best at 76.8 — for Chinese-language products, this is the first open-weight model at parity with the best closed options

Where V4 trails:

• SWE-Pro: 55.4 vs K2.6's 58.6 — real GitHub issue fixing still favors Kimi by a small margin
• MRCR 1M (long-context retrieval): 83.5 vs Opus 4.6's 92.9 — Claude still holds the crown for finding needles in haystacks
• HLE with tools: 48.2 vs K2.6's 54.0

For Easy tier subscribers ($59.90/month), V4-Flash is included at no extra cost. The model upgrade happened automatically — your assistant just got smarter while you were using it. No config changes, no API key rotation, nothing to approve.

For Premium tier users ($79.90/month), V4-Pro (1.6T total, 49B active) is available by bringing your own DeepSeek API key. The API pricing ($1.74/M input, $3.48/M output) is about 21x cheaper than Claude Opus 4.7, which matters if you're running high-volume workloads. V4-Pro takes things further with a Codeforces rating of 3206 — that beats GPT-5.4 on competitive programming benchmarks.

The deepseek-chat deprecation on July 24 is a non-event for InstantClaw users. The transition is already happening transparently. When DeepSeek officially sunsets the old model IDs, your assistant won't even notice.

A 1M-token window means you can try whole-repo or long-document prompts with less hand-chopping. For genuinely hard work, use a thinking level that matches the job (and expect different latency and cost). For anything that would hurt if it were wrong, verify it anyway—better scores on benchmarks are not a guarantee on your specific file.

In short

V4 is a large jump in open-weight quality and in long-context efficiency. On InstantClaw, Easy already uses V4-Flash on the included path, and Premium can use V4-Pro (and the rest of the OpenClaw model catalog) with a bring-your-own key. You do not need a manual migration to stay current on the way we host OpenClaw; the old API model-name sunset matters most if you call DeepSeek's API yourself outside InstantClaw.