LLM Pricing Digest: DeepSeek Flash Gets a Price Tag, Nine New Models Surface
Week of November 3–8, 2026 — data from LLM Price Watch
One actual price change this week, plus a flood of new model listings. Let me walk through what matters.
The One Price Change: DeepSeek V4.1 Flash
Our tracker picked up pricing on deepseek-v4-1-flash — $0.051 per million input tokens, $0.567 per million output tokens.
To put that in practical terms: if you're running a workload that generates 10 million output tokens a month, you're looking at roughly $5.67. That's very cheap for a Flash-tier model, though it sits in familiar territory for DeepSeek's pricing philosophy — aggressive on input, moderate on output relative to the input cost (the output multiplier here is about 11x input, which is on the higher end).
If your app is output-heavy — think code generation, long-form drafting, detailed summaries — that 11x ratio is worth factoring in when comparing against alternatives. For retrieval-augmented workloads where output is short and inputs are large, the $0.051 input rate is genuinely attractive.
The batch variant (deepseek/deepseek-v4.1-flash-1017:batch) also appeared as a new listing this week, so asynchronous batch processing is an option if latency isn't critical for your use case.
Nine New Models Spotted
This was a busy week for new arrivals across providers. Here's what showed up:
DeepSeek
-
deepseek/deepseek-v4.1-flash-1017:batch— batch variant of the flash model above, first seen Nov 8
Z-AI (GLM series)
-
z-ai/glm-5.2:batch— Nov 8 -
z-ai/glm-5.3-flashx-legal— Nov 3, the-legalsuffix suggests domain-specific fine-tuning, which is worth watching if you work in legal tech
Qwen
-
qwen/qwen3.8-27b:batch— Nov 7, a 27B batch model in Qwen's 3.8 generation
MiniMax
-
minimax/minimax-m3:batch— Nov 7
-
google/gemini-omni-1.2-flash— Nov 6 -
google/gemini-omni-1.1-flash— Nov 6
Two Gemini Omni Flash variants appearing the same day suggests Google is running a parallel versioned rollout. No pricing data to report on these yet — they're freshly listed.
xAI (Grok)
-
x-ai/grok-4.7:batch— Nov 4 -
x-ai/grok-4.6:batch— Nov 4
Two consecutive Grok 4.x batch variants dropping together is a little unusual. Could be that xAI is offering multiple checkpoint versions, or these are separate fine-tunes. Worth keeping an eye on once pricing surfaces.
What This Week Actually Means
Honestly, the story isn't dramatic. One model got priced, nine models appeared on the shelf without pricing yet. The batch theme is consistent — almost every new listing this week is a :batch variant, which tells you providers are leaning into async workflows as a cost-reduction path for high-volume users.
If you're evaluating models right now:
- DeepSeek V4.1 Flash is the only one with confirmed numbers this week. If you need a cheap flash-tier model and your workload tolerates the output pricing ratio, it's worth benchmarking.
- Gemini Omni Flash 1.1/1.2 are on my watchlist — once Google prices those, the comparison against existing Flash models will be interesting.
- GLM 5.3 Flash Legal is a niche one, but if you're building anything in legal document processing, track it.
Nothing from Anthropic or OpenAI moved this week on our tracker.
I update this digest every week based on live data. You can track all of this in real time — price history, new model alerts, and provider comparisons — at llmpricewatch.com.
Top comments (0)