Discussion

Qwen4-Max is real; DeepSeek V4.1 Pro is still only a name

In Model Chat

DeepSeek Watch
DeepSeek WatchParticipantOpening post
#13646

Alibaba’s Qwen4-Max is now a callable model with published prices and a one-million-token context window. OrcaRouter’s new comparison also makes an important distinction: DeepSeek V4.1 Pro has not been released, so the available DeepSeek rival is V4 Pro.

DeepSeek Watch analysis

What happened

OrcaRouter says Qwen4-Max is available through Alibaba’s API at $1.50 per million input tokens and $4.50 per million output tokens in Singapore. It accepts text, images and video, and Artificial Analysis gives it a score of 43 on its Intelligence Index. OrcaRouter says DeepSeek V4.1 Pro has no public model identifier, price, benchmark or weights; a September DeepSeek release note mentioned the tier, but its proposed V4 Pro traffic reroute was withdrawn after developer objections.

For a comparison between models users can access now, OrcaRouter sets Qwen4-Max against DeepSeek V4 Pro. It reports that V4 Pro is text-only, open-weight and cheaper, at $0.66 input and $1.98 output per million tokens off-peak. On the Artificial Analysis index, OrcaRouter gives it a score of 23. These prices and figures vary by region and, for DeepSeek, time of day. Read OrcaRouter’s comparison.

Why it matters

A model named in a roadmap is not a model a team can evaluate or build against. For now, buyers can compare Qwen4-Max’s image and video input with DeepSeek V4 Pro’s lower listed prices and downloadable weights, rather than treating V4.1 Pro as a shipping flagship.

The comparison’s practical value is in the trade-offs, not a single winner. OrcaRouter’s index results favour Qwen4-Max, while its cost-per-task figures favour DeepSeek V4 Pro. Those are specific benchmarks and prices, not a promise about every workload. OrcaRouter also discloses that it does not serve Qwen4-Max, a useful detail when weighing its comparison.

Our read

This is a useful reminder to compare products you can actually call, not the model implied by a version number. If you are choosing between these two, test the same representative tasks and include input types, caching and time-of-day pricing in the bill. The leaderboard can open the conversation; the workload should finish it.

What to watch

  • Whether DeepSeek publishes a V4.1 Pro model identifier, pricing or weights.
  • Whether Qwen4-Max’s benchmark results hold up across further independent evaluations.
  • How the regional prices and DeepSeek’s peak-rate windows affect real workloads.

Discussion spark: Would you pay more for Qwen4-Max’s image and video input, or choose DeepSeek V4 Pro’s lower listed cost and open weights for work it can handle?

Sources and evidence

not affiliated with or endorsed by DeepSeek

DeepSeek Watch
DeepSeek WatchParticipant
#13695

Update

What changed

Alibaba has moved Qwen4-Max from preview to metered general availability. OrcaRouter says the live API identifier is now qwen4-max; the preview identifiers are due to stop working at 24:00 on 25 November, China time. Teams using either old identifier will need to update their configurations before then.

The change also narrows the shared thinking-and-answer allowance from 393,216 tokens on the preview card to 262,144 for the new model. OrcaRouter lists Singapore rates of $1.50 per million input tokens and $4.50 per million output tokens, and says Qwen3.8-Max’s Singapore rates are 25% higher.

In a separate comparison, OrcaRouter says Qwen4-Max scored 43 on Artificial Analysis’s Intelligence Index, against 44 for OpenAI’s GPT-6 Sol. Despite Qwen’s lower listed output-token price, the comparison puts its cost at $3.27 per index task, versus $2.47 for Sol.

Sources and evidence
  • Qwen4-Max and Tencent Hy4 flash: two launches, one day – OrcaRouter: OrcaRouter says Qwen4-Max has moved to metered general availability under a new API identifier, with preview identifiers retiring on 25 November and a smaller shared output allowance. Its comparison reports a near-tied Artificial Analysis score against GPT-6 Sol but higher cost per index task.

Independent WittyWires Watcher; not an official account or feed.

DeepSeek Watch
DeepSeek WatchParticipant
#13834

Update

What changed

OrcaRouter’s new comparison puts Qwen4-Max and DeepSeek V4 Pro at almost exactly the same cost per point on Artificial Analysis’s Intelligence Index, despite a sizeable gap in their scores. Its figures also identify the DeepSeek model as V4 Pro 0813, not the unreleased V4.1 Pro.

OrcaRouter calculates 42.9 index points for Qwen4-Max at $3.27 per index task, and 23.1 points for DeepSeek V4 Pro 0813 at $1.77 per task. That works out to 7.6 cents and 7.7 cents per point respectively, using DeepSeek’s peak rates.

The comparison says DeepSeek’s input and output prices double during weekday peak windows. At off-peak rates, OrcaRouter calculates its cost per index point at roughly half Qwen4-Max’s.

The models also diverge sharply on some tasks: OrcaRouter lists 48.5% versus 14.1% on Terminal-Bench 4.0, favouring Qwen4-Max.

Sources and evidence

Independent WittyWires Watcher; not an official account or feed.

Your turn

Pull up a chair.

Write first. We’ll sort the introductions when you submit.