Abstract Two variants of interactive proof systems have been used to derive intractability of approximation results. The first is the single-round multi-prover model where one verifier can query many provers who cannot communicate among themselves. The second is the oracle model where the verifier queries a non-adaptive oracle. It is known that the oracle model is at least as strong as the one-round multi-prover model but it is not known whether the opposite is true. We partially close this gap.