Compare/Sarvam 105B (high) vs Qwen3.5 4B (Non-reasoning)

Sarvam 105B (high)vsQwen3.5 4B (Non-reasoning)

Side-by-side comparison of pricing, 12 benchmarks, and generation speed.

Sarvam

Sarvam 105B (high)

Input
$0.04/M
Output
$0.17/M
Speed
TTFT
Alibaba

Qwen3.5 4B (Non-reasoning)

Input
$0.03/M
Output
$0.15/M
Speed
28 tok/s
TTFT
0.71s

Winner by Category

Cheaper
Qwen3.5 4B (Non-reasoning)
Faster (tok/s)
Qwen3.5 4B (Non-reasoning)
Lower Latency
Qwen3.5 4B (Non-reasoning)
Benchmarks (0-2)
Qwen3.5 4B (Non-reasoning)

Pricing Comparison

MetricSarvam 105B (high)Qwen3.5 4B (Non-reasoning)
Input ($/M tokens)$0.04$0.03
Output ($/M tokens)$0.17$0.15
Cost for 1M input + 100K output tokens:
Sarvam 105B (high)$0.06
Qwen3.5 4B (Non-reasoning)$0.04

Speed Comparison

Output Speed (tokens/s) — higher is better
Sarvam 105B (high)
Qwen3.5 4B (Non-reasoning)
28 tok/s
Time to First Token (seconds) — lower is better
Sarvam 105B (high)
Qwen3.5 4B (Non-reasoning)
0.71s

Editorial Analysis

Verdict. Qwen3.5 4B (Non-reasoning) takes the aggregate benchmark matchup 2–0 across 2 categories. Real workloads usually care about a handful of specific tasks — see the per-benchmark table above.

Pricing. Both models sit in the budget bracket for output-token pricing. At 1.1× the per-million-token cost, Qwen3.5 4B (Non-reasoning) is meaningfully cheaper if your traffic is output-heavy (long completions, document generation, agent loops). Qwen3.5 4B (Non-reasoning) makes more sense when output volume is low and absolute reasoning quality justifies the premium.

Strengths. Sarvam 105B (high) is strongest on Intelligence Index (11.9). Qwen3.5 4B (Non-reasoning) leads on Coding Index (20.3), Intelligence Index (16.1).

Speed. Speed data is incomplete for this pair; benchmark and price should decide.

Provider. Sarvam and Alibaba sell to overlapping but distinct developer audiences: Sarvam tends to ship frontier reasoning models with premium positioning, while Alibaba often prices more aggressively. Your existing vendor relationships, billing, and SLA preferences may matter as much as the raw numbers above.

Workload cost. Workload scenarios (per million requests at 30M input + 15M output tokens): Sarvam 105B (high) costs $3.75 ($45/year); Qwen3.5 4B (Non-reasoning) costs $3.15 ($38/year). At a smaller 5M-input/2M-output scale (single-developer tool or prototype): Sarvam 105B (high) ≈ $0.54/run, Qwen3.5 4B (Non-reasoning) ≈ $0.45/run. At agent/realtime scale (200M input / 100M output per million requests): Sarvam 105B (high) ≈ $25/run, Qwen3.5 4B (Non-reasoning) ≈ $21/run. Qwen3.5 4B (Non-reasoning) becomes more attractive at higher volume — the absolute per-token pricing difference compounds when you ship at scale.

Recommendation. Both models have legitimate use cases — the right answer depends on whether you are optimizing for benchmark ceiling, latency, or unit cost. Start with the cheaper / faster model, evaluate against your specific task, and only switch if the upgrade shows a meaningful lift.

Benchmark Comparison

Data from Artificial Analysis API — 12 benchmarks

Intelligence Index
11.916.1
Coding Index
20.3
Math Index
GPQA Diamond
MMLU-Pro
LiveCodeBench
AIME 2025
MATH-500
Humanity's Last Exam
SciCode
IFBench
TerminalBench
Sarvam 105B (high)0 wins
2 winsQwen3.5 4B (Non-reasoning)

Frequently Asked Questions

Which is cheaper, Sarvam 105B (high) or Qwen3.5 4B (Non-reasoning)?

Qwen3.5 4B (Non-reasoning) is cheaper overall. Its blended price (3:1 input/output ratio) is $0.06/M tokens vs $0.07/M for Sarvam 105B (high).

Which model performs better on benchmarks?

Qwen3.5 4B (Non-reasoning) wins 2 out of 12 benchmarks compared to 0 for Sarvam 105B (high). See the detailed benchmark chart above for per-category results.

Which is faster for real-time applications?

Qwen3.5 4B (Non-reasoning) generates tokens faster at 28 tok/s vs — tok/s. However, Qwen3.5 4B (Non-reasoning) has lower time-to-first-token (0.71s vs —s).

When should I use Sarvam 105B (high) vs Qwen3.5 4B (Non-reasoning)?

Choose based on your priorities: Qwen3.5 4B (Non-reasoning) for lower cost, Qwen3.5 4B (Non-reasoning) for stronger benchmark performance, and Qwen3.5 4B (Non-reasoning) for faster generation. For latency-sensitive apps, check the TTFT comparison above.