Head-to-head showdown: Databricks DBRX Instruct ($0.60 in / $1.80 out per 1M) vs Claude Sonnet 4.6 (OpenRouter) ($3.00 in / $15.00 out per 1M). Databricks DBRX Instruct is 7.5× cheaper across standard token mixes, with 32.8K vs 1M context windows.
databricks/dbrx-instruct · databricks
claude-sonnet-4.6 · anthropic
Benchmark
| Workload Scenario | Databricks DBRX Instruct | Claude Sonnet 4.6 (OpenRouter) | Price Delta |
|---|---|---|---|
| 1M input tokens (raw text) | $0.60 | $3.00 | −$2.40 |
| 1M output tokens (generation) | $1.80 | $15.00 | −$13.20 |
| 1M tokens · 70% input / 30% output mix | $0.96 | $6.60 | −$5.64 |
| Standard chat turn (4K in / 800 out, 50% cached) | $0.00384 | $0.0186 | −$0.0148 |
| Monthly scale (10K requests / day) | $1,152.00 | $5,580.00 | −$4,428.00 |
Negative difference = Databricks DBRX Instruct is cheaper. Positive = Claude Sonnet 4.6 (OpenRouter) is cheaper.
Scaling Curve
Total cost of a token volume at 70/30 input/output split (uncached). Log-log scale.
Verdict
Databricks DBRX Instruct is cheaper on both input and output rates, so it costs less at every input/output mix. Price alone still isn't the whole decision: capability, latency and context limits (Databricks DBRX Instruct: 32.8K, Claude Sonnet 4.6 (OpenRouter): 1M) may justify the premium for your task.
FAQ
On input, Databricks DBRX Instruct is cheaper ($0.60/M vs $3.00/M). On output, Databricks DBRX Instruct is cheaper ($1.80/M vs $15.00/M). The same model is cheaper on both sides, so it wins at every mix.
A chat-style request (4,000 input + 800 output tokens, 50% cached) costs $0.00384 on Databricks DBRX Instruct and $0.0186 on Claude Sonnet 4.6 (OpenRouter) — Databricks DBRX Instruct is 4.8× cheaper for that workload.
Databricks DBRX Instruct supports 32,768 tokens (4.1K max output); Claude Sonnet 4.6 (OpenRouter) supports 1,000,000 (128K max output). Claude Sonnet 4.6 (OpenRouter) fits 30.5× more context, which matters for long documents and agents.