LLM Comparison
Nemotron 3 Super vs Step 3.5 Flash
Side-by-side specs, pricing & capabilities
Add to comparison
2/6 models| Organization | ||
| OpenTools Score | ||
| Family | Nemotron | Step |
| Status | Current | Current |
| Release Date | Mar 2026 | Jan 2026 |
| Context Window | 262K tokens | 262K tokens |
| Input Price | $0.09/M tokens | $0.10/M tokens |
| Output Price | $0.45/M tokens | $0.30/M tokens |
| Capabilities | textcode | textcode |
| Max Output | — | 66K tokens |
| API Identifier | nvidia/nemotron-3-super-120b-a12b | stepfun/step-3.5-flash |
| View Nemotron 3 Super | View Step 3.5 Flash |
Cost Calculator
Enter your expected monthly token usage to compare costs.
| Model | Input | Output | Total / mo | vs Best |
|---|---|---|---|---|
| Step 3.5 FlashCheapest | $0.10 | $0.15 | $0.25 | — |
| Nemotron 3 Super | $0.09 | $0.23 | $0.32 | +26% |
Compare hosting configurations
Compare two exact setups, including different configurations of the same model. Hardware fit, measured speed and costs are separate decisions.
Add your cost assumptions (optional)
Use hours for the same scenario period. Cloud billing includes idle time; active hours alone do not set the bill. Local electricity uses whole-system attributable power, powered hours and your tariff. Missing inputs remain unknown.
Measured speed needs the same pinned workload protocol, tokenizer, context, batching and cache state.
| Configuration | Run full native Nemotron 3 Super reasoning and tools on eight owned H100 GPUs | Host full native Step 3.5 Flash reasoning and tools on an owned eight-H100 server |
|---|---|---|
| Model | Nemotron 3 Super | Step 3.5 Flash |
| Hardware | Owned eight-H100 planning profile | Existing eight-H100 80 GB GPU server |
| Runtime | Original Super full BF16 target plus native two-physical-layer MTP vLLM 0.27.1 / 6e448d0; immutable amd64 image, source-guarded modules; CUDA 13.0.3/Python 3.12 source defaults, installed inventory reader-checked · CUDA | Released native Step3.5 Flash BF16 with three-layer MTP on vLLM vLLM 0.27.1 commit 6e448d0; exact official Linux amd64 image digest, reader preflight required · CUDA |
| Artifact | Original Nemotron 3 Super 120B A12B full BF16 checkpoint with native MTP · Full native BF16 target/MTP; explicit source FP32 Mamba A and routing correction state retained · Full native BF16, including every expert and both physical MTP layers; no quantization | Full native Step 3.5 Flash BF16 with three MTP layers · Complete native BF16 weights, including every expert and MTP layers 45,46,47 · Unquantized native BF16; complete experts and all MTP tensors retained |
| Context / concurrency | 8,192 tokens / 1 | 8,192 tokens / 1 |
| Required memory | Unknown (unknown) | Unknown (unknown) |
| Evidence | Documented setup | Documented setup |
| Hardware fit | Hardware fit unknown No measured per-device or whole-process peak for full BF16 target/MTP/FP32 SSM at8192context. Enter available memory per device or unified memory pool. Host RAM: Nominal2TiB host allocation and1900GiB usable threshold are planning, not measured minima. Free disk space: Chosen1TiB allocation/900GiB free threshold is planning; full runtime/compiler/cache/loading requirements remain unmeasured. | Hardware fit unknown No measured per-device or whole-process peak for full BF16/MTP3 on eight H100s at8192 context. Enter available memory per device or unified memory pool. Host RAM: 2048 GiB host allocation is a planning choice, not measured minimum. Free disk space: 768 GiB allocated disk and600 GiB free-space guard are explicit prescriptions; complete installed runtime/cache requirements remain unmeasured. |
| One-time cost | Unknown hardware: Owned server/GPU acquisition and incremental purchase cost unknown. setup: Reader setup, package/image transfer and labor cost unknown. | Unknown hardware: Existing owned GPU/host purchase and incremental cost are unknown. setup: Reader setup and labor cost is unknown. |
| Recurring scenario cost | Unknown electricity: Actual wall power, local rate and powered hours unknown; GPU TDP is not wall-energy proof. storage: Incremental local 1 TiB storage cost unknown. egress: Actual transfer/allowance and network charges unknown. administration: Maintenance time/rate unknown. One-time setup and recurring costs are separate. Unknown components are not treated as free. No monthly token output is inferred from peak throughput. | Unknown electricity: Actual wall power/rate/poweredhours unknown; no GPU TDP conversion. storage: Incremental owned 768 GiB storage costs are unknown. egress: Actual transfer/allowance/traffic cost unknown. administration: Maintenance time/rate unknown. One-time setup and recurring costs are separate. Unknown components are not treated as free. No monthly token output is inferred from peak throughput. |
| Published measurements (unranked) | No compatible measurement available | No compatible measurement available |
Different currencies and incomplete costs are not ranked. Model quality benchmarks remain on the model pages; they do not establish quality or speed for every deployment.
NVIDIA
Nemotron 3 Super
Nemotron 3 Super is a large language model from NVIDIA. Supports up to 262,144 token context window. Available from $0.09/M input tokens.
StepFun
Step 3.5 Flash
Step 3.5 Flash is a large language model from StepFun. Supports up to 262,144 token context window. Available from $0.10/M input tokens.
More Comparisons
Looking for more AI models?
Browse All LLMs